------------------------------------------------------------------------------------------------------------------
      name:  <unnamed>
       log:  C:\Users\mbeissin\Desktop\Stata files for book\Robustnesstestfiles\Logfiles\robustnesstestschapter4.l
> og
  log type:  text
 opened on:  26 Jan 2022, 10:28:03

. * ============================================================================
. * ROBUSTNESS CHECKS FOR STATISTICAL RESULTS APPEARING IN CHAPTER 4
. * STATA  
. * Robustness checks for results reported in Chapter 4  
. * Author: Mark R. Beissinger  
. * Date:  January 2022  
. * Princeton, NJ 
. * =============================================================================
. * BEFORE RUNNING, YOU MUST SET THE DEFAULT PATH FOR WHERE THE DATA
. *   FILES RESIDE
. * ============================================================================
. * The following datafiles are used in this file:
. *   Dataset of revolutionary episodes--revolutionaryeps.dta
. *   Multiple imputation dataset for revolutionary episodes (regime variables)--
. *               revolutionaryepsmireg.dta
. *       Multiple imputation dataset for revolutionary episodes (opposition variables)--
. *               revolutionaryepsmiopp.dta
. *       Multiple imputation dataset for revolutionary episodes (combined model)--
. *               revolutionaryepsmicomb.dta
. * =============================================================================
. * Before running, you must download the following packages for STATA:
. *       checkrob from http://fmwww.bc.edu/RePEc/bocode/c
. *       collin from https://stats.oarc.ucla.edu/stata/ado/analysis/
. *       eststo and esttab from http://www.stata-journal.com/software/sj7-2
. * =============================================================================
. * The following output is produced by these robustness tests:
. *               Robustnesstestfiles\Logfiles\robustnesstestschapter4.log
. *
. *       The following graphs are produced by these robustness tests:
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat1.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat2.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat3.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat4.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat5.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat6.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat7.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat8.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_1_scat9.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat1.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat2.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat3.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat4.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat5.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat6.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat7.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat8.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_2_scat9.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat1.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat2.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat3.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat4.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat5.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat6.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat7.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat8.pdf
. *               Robustnesstestfiles\Logfiles\robch4tab4_3_scat9.pdf
. *
. *       These files have been combined with the logfile for the chapter into a 
. *               single output file, located in the Robustnesstestfiles\Outputfiles
. *               folder 
. *       In addition, the reworked output from the checkrob procedure run in this
. *          chapter can be viewed in the Excel file checkrob.results.chapter4.xlsx,
. *               also located in the Robustnesstestfiles\Outputfiles folder              
. * =============================================================================
. 
. clear

. use revolutionaryeps.dta

. quietly: logit success c.newpolitymin1##c.newpolitymin1 newincumbpowerdur newgdppcthl newlnoill i.civilwar##c.ne
> wmilexpsold10tile if startyear>1899 & colony==0, or nolog

. generate sample=e(sample)

. 
. * =============================================================
. * ROBUSTNESS TESTS FOR MODELS 6 & 9 IN TABLE 4.1--REGIME MODEL
. * =============================================================
. 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Testing robustness of the specification to inclusion or exclusion of variables
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * ************
. * PLEASE READ
. * ************
. * The checkrob procedure checks all possible combinations of variables and tests 
. *   if signs are stable and whether z-statistics for Beta/S.E. of Beta are >1.96 
. *               (i.e., .05 level of significance) or >1.65 and <1.96 (i.e., .10 level of 
. *               significance).
. * All z-statistics were calculated in Excel from the comma-separated tables 
. *               produced from the output of checkrob.      
. * BEWARE: There is a glitch in the checkrob procedure involving quadratic 
. *               specifications. In the tables produced by the procedure, the results for
. *               the first variable of a quadratic specification sometimes falsely 
. *       include results from the squared variable. ONE MUST VISUALLY INSPECT 
. *               AND, IF PRESENT, HAND-CORRECT.
. * I HAVE PROVIDED AN EXCEL TABLE WITH RESULTS OF THE TEST BELOW THAT CORRECTS
. *       FOR THE ABOVE PROBLEM AND CALCULATES THE Z-SCORES (See the Excel file 
. *       checkrob.results.chapter4.xlsx, which contains all results of all checkrob
. *       tests in this chapter).
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * THE FOLLOWING COMMANDS WERE USED TO CREATE THE EXCEL TABLE.
. * DO NOT RUN THESE COMMANDS UNLESS YOU INTEND TO RECONFIGURE THE OUTPUT
. *       TO CORRECT FOR THE ABOVE GLITCH AND TO CALCULATE THE Z-SCORES BY HAND.
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Recreate common sample for complete-case sample (Model 9)
. *   quietly:  logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newincumbage newgdppcthl newlnoill n
> ewmilexpsold10tile civilwar newcivxmilexp if startyear>1899, or
. *   generate sample=e(sample)
. * Run checkrob command
. *   checkrob 2 6 ch4tab1mod9.txt:  logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl new
> lnoill newmilexpsold10tile civilwar newcivxmilexp if startyear>1899 & sample==1, or
. * RESULTS:
. *       Given that the variables for newpolitymin1 and newpolitymin1sq must be together, there are 32 possible c
> ombinations
. *               (Do note, however, that civilwar, newilexpsold10tile, and newcivxmilexp form an interaction)
. *       --newpolitymin1 and newpolitymin1sq are significant at .05 level or better in 100% of specifications, wi
> th no sign changes
. *   --newincumbpowerdur is significant at .05 level in 100% of specifications, with no sign changes
. *       --newgdppcthl is significant at the .05 level in 37.5%, significant at the .10 level in 6.3%, and insign
> ificant in 56.3% of specifications; 
. *               sign changes:  positive in 53.1% (almost all of which are statistically insignificant, with two 
> significant at the .10 level) and 
. *               negative in 46.9% (and in most cases, statistically significant at the .05 level)
. *               newgdppcthl appears interact with some of the other variables in the specification--particularly
>  newlnoill and newmilexpsold10tile
. *       --newlnoill is significant at the .05 level in 93.8% of specifications and at the .10 level in 6.3% of s
> pecifications, with no sign changes (consistently negative)
. *       --when not in interaction, newmilexpsold10tile is always significant at the .05 level and positive, no s
> ign changes
. *       --when not in interaction, civilwar is always statistically significant and negative;
. *       --in the interaction civilwar x newmilexpsold10tile (8 specifications in which they appear together--all
>  others are ignored):
. *               newmilexpsold10tile is always significant at the .05 level and positive , no sign changes
. *               civilwar in the interaction is never statistically significant, but consistently negative
. *               newcivxmilexp is consistently negative and statistically significant at the .10 level for half o
> f the specifications of interactions and insignficant in 50 percent
. 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Bootstrapped standard errors--1000 replications--complete-case sample
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp if startyear>1899 & sample==1, or nolog vce(bootstrap, bca seed(1234) rep(1000))
(running logit on estimation sample)

Jackknife replications (234)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200
..................................

Bootstrap replications (1000)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200
..................................................   250
..................................................   300
..................................................   350
..................................................   400
..................................................   450
..................................................   500
..................................................   550
..................................................   600
..................................................   650
..................................................   700
..................................................   750
..................................................   800
..................................................   850
..................................................   900
..................................................   950
..................................................  1000

Logistic regression                             Number of obs     =        234
                                                Replications      =      1,000
                                                Wald chi2(8)      =      37.90
                                                Prob > chi2       =     0.0000
Log likelihood =  -118.0399                     Pseudo R2         =     0.2381

-------------------------------------------------------------------------------------
                    |   Observed   Bootstrap                         Normal-based
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9043827   .0335689    -2.71   0.007     .8409251     .972629
    newpolitymin1sq |    .973352   .0074318    -3.54   0.000     .9588944    .9880275
  newincumbpowerdur |   1.063515   .0236503     2.77   0.006     1.018157    1.110894
        newgdppcthl |   .8621668    .075793    -1.69   0.092     .7257085    1.024284
          newlnoill |   .8692383   .0358624    -3.40   0.001     .8017161    .9424473
newmilexpsold10tile |    1.49645    .148471     4.06   0.000     1.231998    1.817668
           civilwar |   1.028174   .8807224     0.03   0.974     .1918389    5.510569
      newcivxmilexp |   .8080672   .1188463    -1.45   0.147     .6056995    1.078047
              _cons |   .3343246   .1700795    -2.15   0.031     .1233503    .9061424
-------------------------------------------------------------------------------------

. estat bootstrap, all

Logistic regression                             Number of obs     =        234
                                                Replications      =       1000

------------------------------------------------------------------------------
             |    Observed               Bootstrap
     success |       Coef.       Bias    Std. Err.  [95% Conf. Interval]
-------------+----------------------------------------------------------------
newpolitym~1 |  -.10050265  -.0055531   .03711806   -.1732527  -.0277526   (N)
             |                                      -.1803383  -.0388121   (P)
             |                                      -.1748927  -.0358668  (BC)
             |                                      -.1695114  -.0333782 (BCa)
newpolitym~q |   -.0270095  -.0017366   .00763524   -.0419743  -.0120447   (N)
             |                                      -.0445673  -.0146911   (P)
             |                                       -.040657  -.0126279  (BC)
             |                                      -.0404018  -.0115693 (BCa)
newincumbp~r |   .06157981   .0036249   .02223784    .0179945   .1051652   (N)
             |                                       .0236071   .1078471   (P)
             |                                       .0186888   .1025739  (BC)
             |                                       .0181834   .1002967 (BCa)
 newgdppcthl |  -.14830658  -.0075479   .08790994   -.3206069   .0239937   (N)
             |                                      -.3363244   .0090078   (P)
             |                                      -.3266475   .0229503  (BC)
             |                                      -.3251484   .0256677 (BCa)
   newlnoill |  -.14013799  -.0073188   .04125726   -.2210007  -.0592752   (N)
             |                                      -.2285454  -.0638661   (P)
             |                                      -.2121324  -.0529835  (BC)
             |                                      -.2110479  -.0465257 (BCa)
newmilexps~e |   .40309578   .0199817   .09921544    .2086371   .5975545   (N)
             |                                       .2427562   .6317849   (P)
             |                                       .2138415   .5856377  (BC)
             |                                       .2106218    .581956 (BCa)
    civilwar |   .02778416  -.0481642   .85658906     -1.6511   1.706668   (N)
             |                                      -1.781158   1.663033   (P)
             |                                      -1.783952   1.641183  (BC)
             |                                      -1.773534   1.701518 (BCa)
newcivxmil~p |  -.21311002  -.0087621   .14707481   -.5013713   .0751513   (N)
             |                                      -.5311711   .0661687   (P)
             |                                      -.5210981   .0774775  (BC)
             |                                      -.5154096   .0832355 (BCa)
       _cons |  -1.0956428  -.0316665   .50872568   -2.092727  -.0985588   (N)
             |                                      -2.195235  -.1481386   (P)
             |                                      -2.131612  -.0902695  (BC)
             |                                      -2.082471  -.0598782 (BCa)
------------------------------------------------------------------------------
(N)    normal confidence interval
(P)    percentile confidence interval
(BC)   bias-corrected confidence interval
(BCa)  bias-corrected and accelerated confidence interval

. *  Result:  no changes in signs or patterns of significance except that
. *                         interaction variable newcivxmilexp turns insignificant
. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Bootstrapping multiple imputation estimation--regime model
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * using 10 imputations, 100 bootstraps (20 imputations takes a very long time)
. * WARNING: THIS TEST CAN TAKE 40 MINUTES TO COMPLETE
. use revolutionaryeps.dta, clear

. drop if colony==1
(57 observations deleted)

. program define myboot, rclass
  1.    mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) 
> newlnoill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmile
> xp = success civilwar, add(10) rseed(1234) force
  2.    mi estimate, eform: logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill ne
> wmilexpsold10tile civilwar newcivxmilexp, or
  3.    return scalar b_npo1 = el(e(b_mi),1,1)
  4.    return scalar b_npsq = el(e(b_mi),1,2)
  5.    return scalar b_newi = el(e(b_mi),1,3)
  6.    return scalar b_newg = el(e(b_mi),1,4)
  7.    return scalar b_newo = el(e(b_mi),1,5)
  8.    return scalar b_newm = el(e(b_mi),1,6)
  9.    return scalar b_civw = el(e(b_mi),1,7)
 10.    return scalar b_cixm = el(e(b_mi),1,8)
 11.    return scalar b_int  = el(e(b_mi),1,9)
 12. end

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newincumbage newgdppcthl newlnoill newmilexp
> sold10tile newcivxmilexp

. set seed 1234

. bootstrap b_newpolitymin1=r(b_npo1) b_newpolitymin1sq=r(b_npsq) b_newincumbpowerdur=r(b_newi) b_newgdppcthl=r(b_
> newg) b_newlnoill=r(b_newo) b_newmilexpsold10tile=r(b_newm) b_civilwar=r(b_civw) b_newcivxmilexp=r(b_cixm) inter
> cept=r(b_int), reps(100) eform : myboot 
(running myboot on estimation sample)

Warning:  Because myboot is not an estimation command or does not set e(sample), bootstrap has no way to
          determine which observations are used in calculating the statistics and so assumes that all
          observations are used.  This means that no observations will be excluded from the resampling because
          of missing values or other reasons.

          If the assumption is not true, press Break, save the data, and drop the observations that are to be
          excluded.  Be sure that the dataset in memory contains only the relevant data.

Bootstrap replications (100)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..............................x.....x.............    50
.................x................................   100

Bootstrap results                               Number of obs     =        288
                                                Replications      =         97

      command:  myboot
b_newpolity~1:  r(b_npo1)
b_newpolity~q:  r(b_npsq)
b_newincumb~r:  r(b_newi)
b_newgdppcthl:  r(b_newg)
  b_newlnoill:  r(b_newo)
b_newmilexp~e:  r(b_newm)
   b_civilwar:  r(b_civw)
b_newcivxmi~p:  r(b_cixm)
    intercept:  r(b_int)

---------------------------------------------------------------------------------------
                      |   Observed   Bootstrap                         Normal-based
                      |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
      b_newpolitymin1 |  -.1114748   .0288713    -3.86   0.000    -.1680614   -.0548881
    b_newpolitymin1sq |  -.0218721   .0051221    -4.27   0.000    -.0319111    -.011833
  b_newincumbpowerdur |   .0444188   .0177107     2.51   0.012     .0097065     .079131
        b_newgdppcthl |  -.1690461   .0799902    -2.11   0.035    -.3258241   -.0122681
          b_newlnoill |  -.1205658   .0363645    -3.32   0.001     -.191839   -.0492926
b_newmilexpsold10tile |   .3826171   .0745348     5.13   0.000     .2365316    .5287027
           b_civilwar |  -.2129329   .7133309    -0.30   0.765    -1.611036     1.18517
      b_newcivxmilexp |  -.1908093   .1153747    -1.65   0.098    -.4169396     .035321
            intercept |  -1.016612   .4121864    -2.47   0.014    -1.824482   -.2087411
---------------------------------------------------------------------------------------
Note: One or more parameters could not be estimated in 3 bootstrap replicates;
      standard-error estimates include only complete replications.

. * Displaying exponentiated form
. bootstrap, eform

Bootstrap results                               Number of obs     =        288
                                                Replications      =         97

      command:  myboot
b_newpolity~1:  r(b_npo1)
b_newpolity~q:  r(b_npsq)
b_newincumb~r:  r(b_newi)
b_newgdppcthl:  r(b_newg)
  b_newlnoill:  r(b_newo)
b_newmilexp~e:  r(b_newm)
   b_civilwar:  r(b_civw)
b_newcivxmi~p:  r(b_cixm)
    intercept:  r(b_int)

---------------------------------------------------------------------------------------
                      |   Observed   Bootstrap                         Normal-based
                      |     exp(b)   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
      b_newpolitymin1 |    .894514   .0258257    -3.86   0.000     .8453019     .946591
    b_newpolitymin1sq |   .9783654   .0050112    -4.27   0.000     .9685927    .9882367
  b_newincumbpowerdur |    1.04542   .0185151     2.51   0.012     1.009754    1.082346
        b_newgdppcthl |   .8444699   .0675494    -2.11   0.035     .7219321    .9878068
          b_newlnoill |   .8864187   .0322342    -3.32   0.001     .8254397    .9519025
b_newmilexpsold10tile |   1.466117   .1092767     5.13   0.000     1.266848     1.69673
           b_civilwar |   .8082104   .5765214    -0.30   0.765     .1996807    3.271243
      b_newcivxmilexp |   .8262901    .095333    -1.65   0.098     .6590607    1.035952
            intercept |   .3618189   .1491368    -2.47   0.014     .1613012    .8116053
---------------------------------------------------------------------------------------
Note: One or more parameters could not be estimated in 3 bootstrap replicates;
      standard-error estimates include only complete replications.

. clear programs

. *  Result:  no changes in signs or patterns of significance
. 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Testing for multicollinearity in complete case sample 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++
. use revolutionaryeps.dta, clear

. quietly: logit success c.newpolitymin1##c.newpolitymin1 newincumbpowerdur newgdppcthl newlnoill i.civilwar##c.ne
> wmilexpsold10tile if startyear>1899 & colony==0, or nolog

. generate sample=e(sample)

. collin newpolitymin1  newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar newci
> vxmilexp if startyear>1899 & sample==1
(obs=234)

  Collinearity Diagnostics

                        SQRT                   R-
  Variable      VIF     VIF    Tolerance    Squared
----------------------------------------------------
newpolitymin1      1.15    1.07    0.8662      0.1338
newpolitymin1sq      1.08    1.04    0.9237      0.0763
newincumbpowerdur      1.17    1.08    0.8538      0.1462
newgdppcthl      1.89    1.38    0.5279      0.4721
 newlnoill      1.16    1.08    0.8632      0.1368
newmilexpsold10tile      2.46    1.57    0.4065      0.5935
  civilwar      5.39    2.32    0.1854      0.8146
newcivxmilexp      6.00    2.45    0.1667      0.8333
----------------------------------------------------
  Mean VIF      2.54

                           Cond
        Eigenval          Index
---------------------------------
    1     5.3024          1.0000
    2     1.2556          2.0550
    3     0.9698          2.3382
    4     0.4717          3.3527
    5     0.4124          3.5856
    6     0.3011          4.1964
    7     0.1341          6.2889
    8     0.1180          6.7024
    9     0.0348         12.3419
---------------------------------
 Condition Number        12.3419 
 Eigenvalues & Cond Index computed from scaled raw sscp (w/ intercept)
 Det(correlation matrix)    0.0676

. *       RESULT:  VIF all within acceptable range (VIF>5 is cause of concern, VIF>10 indicates significant proble
> m)
. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Visual inspection of potential outliers, complete case model
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. quietly: logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile
>  civilwar newcivxmilexp if startyear>1899 & sample==1, or nolog

. predict pr , pr
(61 missing values generated)

. predict stdres, rstand
(111 missing values generated)

. predict dev, dev
(111 missing values generated)

. predict hat, hat
(111 missing values generated)

. predict dx2, dx2
(111 missing values generated)

. predict dd, dd
(111 missing values generated)

. * Standardized Pearson residuals by predicted probability
. scatter stdres pr, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat1.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat1.pdf written in PDF format)

. *       RESULT: Identified revid 269 (1990 Mongolian Revolution) as potential outlier
. *               Identified revid 47 (Spanish Civil War) as potential outlier
. *               Identified revid 243 (First Chechen War)
. *               Identified revid 399 (Burkinabe Uprising 2014)
. * Standardized Pearson residuals by revid
. scatter stdres revid, mlab(revi)  yline(0) 

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat2.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat2.pdf written in PDF format)

. *       RESULT: Identified revid 269 (1990 Mongolian Revolution) as potential outlier
. *               Identified revid 47 (Spanish Civil War) as potential outlier
. *               Identified revid 243 (First Chechen War)
. * Deviance residual by revid
. scatter dev revid, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat3.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat3.pdf written in PDF format)

. *       RESULT: Identified revid 269 (1990 Mongolian Revolution) as potential outlier
. *               Identified revid 47 (Spanish Civil War) as potential outlier
. *               Identified revid 243 (First Chechen War)
. * Leverage by predicted probability
. scatter hat pr, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat4.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat4.pdf written in PDF format)

. *       RESULT: Identified revid 111 (First Intifada)
. *               Identified revid 399 (Burkinabe Uprising 2014)
. *               Identified revid 135 (Second Intifada)
. *               Identified revid 338 (2006 Hungarian Protests)
. *               Identified revid 399 (Burkinabe Uprising 2014)
. * Leverage by revid (with cutoff point of 3 * the mean of leverage)--easier to see
. mean hat

Mean estimation                   Number of obs   =        234

--------------------------------------------------------------
             |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
         hat |   .0386131   .0013671      .0359197    .0413064
--------------------------------------------------------------

. matrix coefs = e(b)

. local hatmean = 3 * (coefs[1,1])

. scatter hat revid, mlab(revid) yline(0) yline(`hatmean')

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat5.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat5.pdf written in PDF format)

. *       RESULT: Identified revid 111 (First Intifada)
. *               Identified revid 117 (East German Revolution)
. *               Identified revid 152 (Armenian colored revolution attempt)
. * Difference of chi-squares
. scatter dx2 revid, mlab(revid)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat6.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat6.pdf written in PDF format)

. *       RESULT: Identified revid 243 (First Chechen War)
. *               Identified revid 47 (Spanish Civil War) as potential outlier
. * Difference of deviances
. scatter dd revid, mlab(revid)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat7.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat7.pdf written in PDF format)

. *       RESULT: Identified revid 243 (First Chechen War)
. *               Identified revid 47 (Spanish Civil War) out potential outlier
. * Drop predictions
. drop pr stdres dev hat dx2 dd

. 
. * Testing the effect of dropping potential outliers on results in Model 9
. eststo: quietly: logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpso
> ld10tile civilwar newcivxmilexp if startyear>1899 & sample==1, or nolog
(est1 stored)

. eststo: quietly: logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpso
> ld10tile civilwar newcivxmilexp if startyear>1899 & sample==1 & revid!=269 & revid!=47 & revid!=243 & revid!=399
>  & revid!=111 & revid!=135 & revid!=338 & revid!=117 & revid!=152, or nolog
(est2 stored)

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(All_n No_outl)

--------------------------------------------
                      (1)             (2)   
                    All_n         No_outl   
--------------------------------------------
success                                     
newpolitym~1       -0.101**        -0.102** 
                  (-2.91)         (-2.62)   

newpolitym~q      -0.0270***      -0.0269***
                  (-3.81)         (-3.44)   

newincumbp~r       0.0616**        0.0622** 
                   (3.09)          (3.03)   

newgdppcthl        -0.148+         -0.207*  
                  (-1.81)         (-2.17)   

newlnoill          -0.140***       -0.171***
                  (-3.56)         (-3.90)   

newmilexps~e        0.403***        0.485***
                   (4.29)          (4.70)   

civilwar           0.0278           0.101   
                   (0.04)          (0.13)   

newcivxmil~p       -0.213+         -0.293*  
                  (-1.69)         (-2.15)   

_cons              -1.096*         -1.155*  
                  (-2.20)         (-2.24)   
--------------------------------------------
N                     234             225   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  No change in signs and patterns of significance, but newgdppcthl and 
. *               the newcivxmilexp interaction becomes significant at the .05 level
. eststo clear

. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Visual inspection of potential outliers, multiple imputation model
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. use revolutionaryepsmireg.dta, clear

. quietly: mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newinc
> umbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp if startyear>1899

. mi predict xb using miest , xb
(57 missing values generated)

. mi predict stdp using miest, stdp
(57 missing values generated)

. scatter  stdp xb, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat8.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat8.pdf written in PDF format)

. *       RESULT: Identified revid 243 (First Chechen War)
. *                       Identified revid 290 (First Sudanese Civil War) as potential outlier
. *                       Identified revid 331 (Congo Crisis) as potential outlier
. *                       Identified revid 301 (2007 Guinean General Strike) as potential outlier
. *                       Identified revid 259 (228 Uprising) as potential outlier
. *                       Identified revid 298 (Black Friday in Maldives) as potential outlier
. scatter  stdp revid, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_1_scat9.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_1_scat9.pdf written in PDF format)

. *                       Identified revid 383 (Christmas Uprising) as potential outlier
. * Drop predictions
. drop xb stdp

. * Testing the effect of dropping potential outliers on results in Model 6
. eststo: quietly: mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1s
> q newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp if startyear>1899, or
(est1 stored)

. eststo: quietly: mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1s
> q newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp if startyear>1899 & revid~=
> 290 & revid~=331 & revid~=301 & revid~=259 & revid~=298 & revid~=383, or
(est2 stored)

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(All_n No_outl) 

--------------------------------------------
                      (1)             (2)   
                    All_n         No_outl   
--------------------------------------------
success                                     
newpolitym~1       -0.110***       -0.110***
                  (-3.48)         (-3.43)   

newpolitym~q      -0.0218***      -0.0234***
                  (-3.41)         (-3.71)   

newincumbp~r       0.0445*         0.0541** 
                   (2.55)          (2.99)   

newgdppcthl        -0.162*         -0.169*  
                  (-2.07)         (-2.17)   

newlnoill          -0.120***       -0.133***
                  (-3.32)         (-3.62)   

newmilexps~e        0.384***        0.388***
                   (4.15)          (4.33)   

civilwar           0.0311         -0.0164   
                   (0.04)         (-0.02)   

newcivxmil~p       -0.226+         -0.231+  
                  (-1.85)         (-1.88)   

_cons              -1.049*         -0.937*  
                  (-2.26)         (-2.02)   
--------------------------------------------
N                     288             282   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  No changes in signs and patterns of significance
. eststo clear

. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Linktest for omitted variable bias--complete case sample
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. use revolutionaryeps.dta, clear

. quietly: logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile
>  civilwar newcivxmilexp if startyear > 1899 & colony==0, or nolog

. linktest , nolog

Logistic regression                             Number of obs     =        234
                                                LR chi2(2)        =      74.23
                                                Prob > chi2       =     0.0000
Log likelihood = -117.81797                     Pseudo R2         =     0.2396

------------------------------------------------------------------------------
     success |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        _hat |   1.064914   .1800909     5.91   0.000     .7119419    1.417885
      _hatsq |   .0578994   .0844479     0.69   0.493    -.1076155    .2234142
       _cons |  -.0540126   .1855351    -0.29   0.771    -.4176546    .3096295
------------------------------------------------------------------------------

. *       RESULT:  passes, _hatsq is not statistically significant
. 
. 
. * +++++++++++++++++++++++++++++++++++++++++++
. * ROBUSTNESS TESTS FOR MODEL 7 IN TABLE 4.2
. * +++++++++++++++++++++++++++++++++++++++++++
. 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Checkrob procedure testing effect of dropping variables from specification
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * checkrob procedure checks all possible combinations of variables
. * Testing if signs are stable and z-statistics for Beta/S.E. of Beta are >1.96 (i.e., .05 level of significance)
. *   z-statistics were calculated in Excel (as comma-separated tables) from the output produced by checkrob 
. *      .10 level of significance>1.65 and <1.96
. * BEWARE: there is sometimes a glitch in the checkrob procedure
. *    In the tables produced, the results for the first variable of a quadratic specification sometimes falsely 
. *       include some results from the squared variable    
. *    ONE MUST VISUALLY INSPECT AND, IF PRESENT, HAND-CORRECT AND RECALCULATE RESULTS FOR THOSE VARIABLES
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * I HAVE PROVIDED AN EXCEL TABLE WITH RESULTS OF THE TEST BELOW, CORRECTED FOR THE ABOVE PROBLEMS
. *       SEE FILE checkrob.results.chapter4.xlsx
. * THE FOLLOWING COMMANDS WERE USED TO CREATE THE TABLE
. *       (DO NOT RUN UNLESS YOU INTEND TO RECONFIGURE THE OUTPUT BY HAND)
. * Recreate common sample for complete-case sample (Model 9)
. *   quietly:  logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newincumbage newgdppcthl newlnoill n
> ewmilexpsold10tile civilwar newcivxmilexp if startyear>1899, or
. *   generate sample=e(sample)
. * clear
. * use revolutionaryepsmiopp.dta
. * checkrob procedure using Model 6
. * Run checkrob
. * checkrob 0 6 ch4tab2mod7.txt: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimona
> rch if startyear>1899, or
. * RESULTS
. * * RESULTS:
. *       Given that urbandum, deathtile10, and urbxdeathtile10 are in an interaction, the urbxdeathtile10 variabl
> e cannot appear alone
. *               --lnparticnum is significant at the .05 level or better, with no sign changes, for all specifica
> tion
. *               --urbandum generally is statistically significant at the .05 level when it is in interaction wit
> h deathtile10, with no sign changes,
. *                               but does not hold up on its own
. *               --deathtile10 is generally statistically significant at the .05 level when it is in interaction 
> with urbandum, with no sign changes,
. *                               but does not hold up on its own
. *               --the interaction term urbxdeathtile10 is always significant when the other terms in the interac
> tion are included
. *       democrat is significant in all specifications, with no sign changes
. *       antimonarch is generally significant when democrat is included, with no sign changes (but is not signifi
> cant on its own)
. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Bootstrapped standard errors--complete case sample--opposition model
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. use revolutionaryepsmiopp.dta, clear

. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch if startyear>1899, or nolog 
> vce(bootstrap, bca seed(1234) rep(1000)) iterate(15)
(running logit on estimation sample)

Jackknife replications (304)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200
..................................................   250
..................................................   300
....

Bootstrap replications (1000)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200
..................................................   250
..................................................   300
..................................................   350
..................................................   400
..................................................   450
..................................................   500
..................................................   550
..................................................   600
..................................................   650
..................................................   700
..................................................   750
..................................................   800
..................................................   850
..................................................   900
..................................................   950
..................................................  1000

Logistic regression                             Number of obs     =        304
                                                Replications      =      1,000
                                                Wald chi2(6)      =      37.32
                                                Prob > chi2       =     0.0000
Log likelihood =  -171.5114                     Pseudo R2         =     0.1404

---------------------------------------------------------------------------------
                |   Observed   Bootstrap                         Normal-based
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.334822   .1288792     2.99   0.003     1.104685    1.612902
       urbandum |     11.495   13.44991     2.09   0.037     1.160239    113.8861
    deathtile10 |   1.281575   .1607749     1.98   0.048     1.002213    1.638809
urbxdeathtile10 |   .6390092   .0976701    -2.93   0.003     .4735919    .8622038
       democrat |   2.510177   1.007744     2.29   0.022     1.142821    5.513541
    antimonarch |   2.200863    1.05423     1.65   0.100      .860719    5.627621
          _cons |   .0031396   .0039361    -4.60   0.000      .000269    .0366462
---------------------------------------------------------------------------------

. estat bootstrap, all

Logistic regression                             Number of obs     =        304
                                                Replications      =       1000

------------------------------------------------------------------------------
             |    Observed               Bootstrap
     success |       Coef.       Bias    Std. Err.  [95% Conf. Interval]
-------------+----------------------------------------------------------------
 lnparticnum |   .28879759   .0072737   .09655161    .0995599   .4780353   (N)
             |                                       .1078641   .4852535   (P)
             |                                       .1068131   .4833549  (BC)
             |                                       .1019519   .4788014 (BCa)
    urbandum |   2.4419126   .1184781   1.1700657     .148626   4.735199   (N)
             |                                       .3987371    4.98353   (P)
             |                                       .2065202   4.648817  (BC)
             |                                        .087973   4.514083 (BCa)
 deathtile10 |   .24809009   .0127273   .12545102    .0022106   .4939696   (N)
             |                                       .0267034    .529164   (P)
             |                                       .0076924   .5062148  (BC)
             |                                      -.0057338   .4927977 (BCa)
urbxdeath~10 |  -.44783644  -.0235264   .15284609   -.7474093  -.1482636   (N)
             |                                       -.767286  -.1717971   (P)
             |                                       -.740145  -.1406444  (BC)
             |                                      -.7265702  -.1324795 (BCa)
    democrat |   .92035335   .0238037   .40146332    .1334997   1.707207   (N)
             |                                       .1928988   1.762951   (P)
             |                                       .1500418   1.733472  (BC)
             |                                       .1500418   1.733472 (BCa)
 antimonarch |   .78884975   .0388115   .47900727   -.1499872   1.727687   (N)
             |                                      -.0789865   1.723621   (P)
             |                                      -.1873607     1.6656  (BC)
             |                                      -.2054057   1.637646 (BCa)
       _cons |  -5.7636555  -.1867903   1.2537017   -8.220866  -3.306445   (N)
             |                                      -8.758467  -3.610584   (P)
             |                                      -8.391581  -3.463489  (BC)
             |                                      -8.312248  -3.378974 (BCa)
------------------------------------------------------------------------------
(N)    normal confidence interval
(P)    percentile confidence interval
(BC)   bias-corrected confidence interval
(BCa)  bias-corrected and accelerated confidence interval

. *  Result:  all variables significant at the .05 level or better, with exception of antimonarch, which is signif
> icant at the .10 level. 
. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Bootstrapping multiple imputation estimation--opposition model
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * using 10 imputations, 200 bootstraps (20 imputations takes a very long time)
. use revolutionaryeps.dta

. drop if startyear<1900
(2 observations deleted)

. program define myboot, rclass
  1.    mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathti
> le10 = success urbandum democrat antimonarch, add(10) rseed(1234) force
  2.    mi estimate, eform: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch, o
> r
  3.    return scalar b_lnp = el(e(b_mi),1,1)
  4.    return scalar b_urb = el(e(b_mi),1,2)
  5.    return scalar b_d10 = el(e(b_mi),1,3)
  6.    return scalar b_uxd = el(e(b_mi),1,4)
  7.    return scalar b_dem = el(e(b_mi),1,5)
  8.    return scalar b_mon = el(e(b_mi),1,6)
  9.    return scalar b_int  = el(e(b_mi),1,7)
 10. end

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10

. set seed 1234

. bootstrap b_lnparticnum=r(b_lnp) b_urbandum=r(b_urb) b_deathtile10=r(b_d10) b_urbxdeathtile10=r(b_uxd) b_democra
> t=r(b_dem) b_antimonarch=r(b_mon) intercept=r(b_int), reps(100): myboot 
(running myboot on estimation sample)

Warning:  Because myboot is not an estimation command or does not set e(sample), bootstrap has no way to
          determine which observations are used in calculating the statistics and so assumes that all
          observations are used.  This means that no observations will be excluded from the resampling because
          of missing values or other reasons.

          If the assumption is not true, press Break, save the data, and drop the observations that are to be
          excluded.  Be sure that the dataset in memory contains only the relevant data.

Bootstrap replications (100)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100

Bootstrap results                               Number of obs     =        343
                                                Replications      =        100

      command:  myboot
b_lnparticnum:  r(b_lnp)
   b_urbandum:  r(b_urb)
b_deathtile10:  r(b_d10)
b_urbxdeat~10:  r(b_uxd)
   b_democrat:  r(b_dem)
b_antimonarch:  r(b_mon)
    intercept:  r(b_int)

-----------------------------------------------------------------------------------
                  |   Observed   Bootstrap                         Normal-based
                  |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
    b_lnparticnum |   .2940023   .0993039     2.96   0.003     .0993702    .4886343
       b_urbandum |   2.907285   1.114384     2.61   0.009     .7231323    5.091438
    b_deathtile10 |   .2811206   .1266463     2.22   0.026     .0328984    .5293429
b_urbxdeathtile10 |  -.5097852   .1415317    -3.60   0.000    -.7871822   -.2323882
       b_democrat |   1.036911   .4315402     2.40   0.016     .1911077    1.882714
    b_antimonarch |   .9137496   .4791255     1.91   0.057    -.0253192    1.852818
        intercept |  -6.207352    1.26053    -4.92   0.000    -8.677945   -3.736759
-----------------------------------------------------------------------------------

. * Displaying exponentiated form
. bootstrap, eform

Bootstrap results                               Number of obs     =        343
                                                Replications      =        100

      command:  myboot
b_lnparticnum:  r(b_lnp)
   b_urbandum:  r(b_urb)
b_deathtile10:  r(b_d10)
b_urbxdeat~10:  r(b_uxd)
   b_democrat:  r(b_dem)
b_antimonarch:  r(b_mon)
    intercept:  r(b_int)

-----------------------------------------------------------------------------------
                  |   Observed   Bootstrap                         Normal-based
                  |     exp(b)   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
    b_lnparticnum |   1.341787   .1332447     2.96   0.003     1.104475    1.630088
       b_urbandum |   18.30703   20.40106     2.61   0.009     2.060878    162.6235
    b_deathtile10 |   1.324613   .1677574     2.22   0.026     1.033445    1.697816
b_urbxdeathtile10 |   .6006246   .0850074    -3.60   0.000     .4551255    .7926383
       b_democrat |   2.820491   1.217155     2.40   0.016      1.21059    6.571317
    b_antimonarch |   2.493655   1.194774     1.91   0.057     .9749986    6.377769
        intercept |   .0020146   .0025394    -4.92   0.000     .0001703    .0238312
-----------------------------------------------------------------------------------

. clear programs

. *  Result:  all variables significant at the .05 level or better, with exception of antimonarch, which is signif
> icant at the .10 level. 
. 
. * ++++++++++++++++++++++++++++++++++
. * Testing for multicollinearity
. * ++++++++++++++++++++++++++++++++++
. use revolutionaryepsmiopp.dta, clear

. collin lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch if startyear>1899
(obs=304)

  Collinearity Diagnostics

                        SQRT                   R-
  Variable      VIF     VIF    Tolerance    Squared
----------------------------------------------------
lnparticnum      1.52    1.23    0.6573      0.3427
  urbandum     11.83    3.44    0.0845      0.9155
deathtile10      4.74    2.18    0.2110      0.7890
urbxdeathtile10      6.31    2.51    0.1585      0.8415
  democrat      1.67    1.29    0.5999      0.4001
antimonarch      1.09    1.05    0.9156      0.0844
----------------------------------------------------
  Mean VIF      4.53

                           Cond
        Eigenval          Index
---------------------------------
    1     4.3870          1.0000
    2     0.9961          2.0986
    3     0.9047          2.2021
    4     0.5238          2.8942
    5     0.1619          5.2047
    6     0.0147         17.2564
    7     0.0118         19.2906
---------------------------------
 Condition Number        19.2906 
 Eigenvalues & Cond Index computed from scaled raw sscp (w/ intercept)
 Det(correlation matrix)    0.0445

. *       RESULT:  urbandum has high VIF (>10) due to presence of interaction variable urbxdeathtile10 
. * Rerun without the interaction term urbxdeathtile10
. collin lnparticnum urbandum deathtile10  democrat antimonarch if startyear>1899
(obs=304)

  Collinearity Diagnostics

                        SQRT                   R-
  Variable      VIF     VIF    Tolerance    Squared
----------------------------------------------------
lnparticnum      1.43    1.20    0.6972      0.3028
  urbandum      2.18    1.48    0.4590      0.5410
deathtile10      1.81    1.34    0.5530      0.4470
  democrat      1.64    1.28    0.6084      0.3916
antimonarch      1.09    1.04    0.9176      0.0824
----------------------------------------------------
  Mean VIF      1.63

                           Cond
        Eigenval          Index
---------------------------------
    1     3.8107          1.0000
    2     0.9850          1.9669
    3     0.8718          2.0906
    4     0.2645          3.7955
    5     0.0561          8.2407
    6     0.0119         17.9195
---------------------------------
 Condition Number        17.9195 
 Eigenvalues & Cond Index computed from scaled raw sscp (w/ intercept)
 Det(correlation matrix)    0.2808

. *       RESULT:  all variables within acceptable range
. 
. * ++++++++++++++++++++++++++++++++++
. * Visual inspection of potential outliers
. * ++++++++++++++++++++++++++++++++++
. *  Create predictions
. use revolutionaryepsmiopp.dta, clear

. quietly: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch if startyear>1899 ,
>  or nolog

. predict pr , pr
(39 missing values generated)

. predict stdres, rstand
(41 missing values generated)

. predict dev, dev
(41 missing values generated)

. predict hat, hat
(41 missing values generated)

. predict dx2, dx2
(41 missing values generated)

. predict dd, dd
(41 missing values generated)

. * Standardized Pearson residuals by predicted probability
. scatter stdres pr, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat1.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat1.pdf written in PDF format)

. *       RESULT: revid 359 (Tunisian independence movement) as potential outlier
. *                       revid 278 (Niger Constitutional Crisis) as potential outlier
. *                       revid 172 (Djiboutian Uprising 2011) as potential outlier
. * Standardized Pearson residuals by revid
. scatter stdres revid, mlab(revi)  yline(0) 

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat2.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat2.pdf written in PDF format)

. *       RESULT: Identified revid 359 (Tunisian independence movement) as potential outlier
. *                       revid 145 (Azerbaijan color revolution attempt) as potential outlier 
. *                       revid 172 (Djiboutian Uprising 2011) as potential outlier
. *                       revid 106 (Solidarity Uprising) as potential outlier
. *                       revid 278 (Niger Constitutional Crisis) as potential outlier
. *                       revid 15 (Irish War of Independence) as potential outlier
. *                       revid 100 (Saur [April] Revolution) as potential outlier
. * Deviance residual by revid
. scatter dev revid, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat3.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat3.pdf written in PDF format)

. *       RESULT: revid 15 (Irish War of Independence) as potential outlier
. *                       revid 100 (Saur Revolution) as potential outlier
. * Leverage by predicted probability
. scatter hat pr, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat4.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat4.pdf written in PDF format)

. *       RESULT: not readable
. * Leverage by revid (with cutoff point of 3 * the mean of leverage)
. mean hat

Mean estimation                   Number of obs   =        304

--------------------------------------------------------------
             |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
         hat |    .033815   .0012443      .0313664    .0362636
--------------------------------------------------------------

. matrix coefs = e(b)

. local hatmean = 3 * (coefs[1,1])

. scatter hat revid, mlab(revid) yline(0) yline(`hatmean')

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat5.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat5.pdf written in PDF format)

. *       RESULT: not readable
. * Difference of chi-squares
. scatter dx2 revid, mlab(revid)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat6.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat6.pdf written in PDF format)

. *       RESULT:  revid 359 (Tunisian independence movement) as potential outlier
. *                        revid 15 (Irish War of Independence) as potential outlier
. *                        revid 100 (Saur Revolution) as potential outlier
. * Difference of deviances
. scatter dd revid, mlab(revid)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat7.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat7.pdf written in PDF format)

. *       RESULT: revid 15 (Irish War of Independence) as potential outlier 
. *                       revid 100 (Saur Revolution) as potential outlier
. *                       revid 278 (Niger Constitutional Crisis) as potential outlier
. *                       revid 172 (Djiboutian Uprising 2011) as potential outlier
. *                       revid 145 (Azerbaijan color revolution attempt) as potential outlier
. *                       revid 155 (Moldovan Twitter Revolution) as potential outlier
. * Drop predictions
. drop pr stdres dev hat dx2 dd

. 
. * Checking to see if dropping potential outliers alters any findings 
. eststo: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch if startyear>1899 , 
> or nolog

Logistic regression                             Number of obs     =        304
                                                LR chi2(6)        =      56.02
                                                Prob > chi2       =     0.0000
Log likelihood =  -171.5114                     Pseudo R2         =     0.1404

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.334822   .1212308     3.18   0.001     1.117161     1.59489
       urbandum |     11.495    11.7528     2.39   0.017     1.549594    85.27081
    deathtile10 |   1.281575   .1453508     2.19   0.029     1.026135    1.600604
urbxdeathtile10 |   .6390092   .0865314    -3.31   0.001     .4900514    .8332447
       democrat |   2.510177   .9234573     2.50   0.012     1.220562    5.162367
    antimonarch |   2.200863   1.072678     1.62   0.106      .846694    5.720839
          _cons |   .0031396   .0034575    -5.23   0.000     .0003627    .0271803
---------------------------------------------------------------------------------
(est1 stored)

. eststo: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch if startyear>1899 & 
> revid~=15 & revid~=100 & revid~=106 & revid~=145 & revid~=155 & revid~=172 & revid~=278 & revid~=359, or nolog

Logistic regression                             Number of obs     =        296
                                                LR chi2(6)        =      70.05
                                                Prob > chi2       =     0.0000
Log likelihood = -159.19916                     Pseudo R2         =     0.1803

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.392835   .1352003     3.41   0.001     1.151529    1.684707
       urbandum |   16.48695   17.77164     2.60   0.009     1.993479    136.3543
    deathtile10 |   1.327554   .1579687     2.38   0.017     1.051394    1.676249
urbxdeathtile10 |   .5840903    .083711    -3.75   0.000     .4410491    .7735226
       democrat |   3.350435   1.285989     3.15   0.002     1.579016    7.109119
    antimonarch |   2.396044   1.198448     1.75   0.081     .8989647    6.386267
          _cons |   .0014755   .0017427    -5.52   0.000     .0001458     .014937
---------------------------------------------------------------------------------
(est2 stored)

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(All_n No_outl)

--------------------------------------------
                      (1)             (2)   
                    All_n         No_outl   
--------------------------------------------
success                                     
lnparticnum         0.289**         0.331***
                   (3.18)          (3.41)   

urbandum            2.442*          2.803** 
                   (2.39)          (2.60)   

deathtile10         0.248*          0.283*  
                   (2.19)          (2.38)   

urbxdeath~10       -0.448***       -0.538***
                  (-3.31)         (-3.75)   

democrat            0.920*          1.209** 
                   (2.50)          (3.15)   

antimonarch         0.789           0.874+  
                   (1.62)          (1.75)   

_cons              -5.764***       -6.519***
                  (-5.23)         (-5.52)   
--------------------------------------------
N                     304             296   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  antimonarch becomes marginally significant, but otherwise no changes in signs or patterns of st
> atistical significance
. eststo clear

. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Visual inspection of potential outliers, multiple imputation model
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. quietly: mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urb
> xdeathtile10 democrat antimonarch if startyear>1899

. mi predict xb using miest , xb

. mi predict stdp using miest, stdp

. scatter  stdp xb, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat8.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat8.pdf written in PDF format)

. *       RESULT: revid 19 (Aster Revolution) as potential outlier
. *                       revid 337 (Argentinian Revolution of 1905) as potential outlier
. *                       revid 250 (1926 Indonesian Communist Revolt) as potential outlier
. *                       revid 61 (La Violencia) as potential outlier
. *                       revid 364 (Tupamaros) as potential outlier
. *                       revid 389 (Ar-Rashid Revolt) as potential outlier
. *                       revid 305 (Vaccine Revolt) as potential outlier
. scatter  stdp revid, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_2_scat9.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_2_scat9.pdf written in PDF format)

. *       RESULT: revid 19 (Aster Revolution) as potential outlier
. *                       revid 337 (Argentinian Revolution of 1905) as potential outlier
. *                       revid 250 (1926 Indonesian Communist Revolt) as potential outlier
. *                       revid 61 (La Violencia) as potential outlier
. *                       revid 364 (Tupamaros) as potential outlier
. *                       revid 389 (Ar-Rashid Revolt) as potential outlier
. *                       revid 305 (Vaccine Revolt) as potential outlier
. * Drop predictions
. drop xb stdp

. * Testing the effect of dropping potential outliers on results in Model 6
. eststo: quietly: mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathti
> le10 urbxdeathtile10 democrat antimonarch if startyear>1899 , or 
(est1 stored)

. eststo: quietly: mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathti
> le10 urbxdeathtile10 democrat antimonarch if startyear>1899 & revid~=19 & revid~=61 & revid~=250 & revid~=305 & 
> revid~=337 & revid~=364 & revid~=389, or
(est2 stored)

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(All_n No_outl) 

--------------------------------------------
                      (1)             (2)   
                    All_n         No_outl   
--------------------------------------------
success                                     
lnparticnum         0.303***        0.296***
                   (3.43)          (3.33)   

urbandum            2.962**         2.993** 
                   (2.87)          (2.90)   

deathtile10         0.281*          0.283*  
                   (2.41)          (2.43)   

urbxdeath~10       -0.520***       -0.515***
                  (-3.81)         (-3.76)   

democrat            1.005**         1.001** 
                   (2.86)          (2.84)   

antimonarch         0.882+          0.802+  
                   (1.95)          (1.76)   

_cons              -6.290***       -6.233***
                  (-5.72)         (-5.65)   
--------------------------------------------
N                     343             336   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  No changes in signs and patterns of significance
. eststo clear

. 
. * ++++++++++++++++++++++++++++++++++
. * Linktest for omitted variable bias
. * ++++++++++++++++++++++++++++++++++
. quietly: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch if startyear>1899, 
> or nolog

. linktest , nolog

Logistic regression                             Number of obs     =        304
                                                LR chi2(2)        =      56.02
                                                Prob > chi2       =     0.0000
Log likelihood = -171.51053                     Pseudo R2         =     0.1404

------------------------------------------------------------------------------
     success |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        _hat |   1.005612   .2022414     4.97   0.000     .6092265    1.401998
      _hatsq |   .0055366   .1329291     0.04   0.967    -.2549997    .2660728
       _cons |  -.0027974   .1626925    -0.02   0.986    -.3216689    .3160741
------------------------------------------------------------------------------

. *       RESULT:  passes, _hatsq is not statistically significant
. 
. 
. * =======================================================================
. * Bivariate relationships between regime characteristics and opposition 
. *       features in the combined model
. * =======================================================================
. * The purpose of these tests is to show how to combining the two models
. *       has an artificiality to it, given the relationships between the
. *       variables in the two models
. 
. clear

. use revolutionaryepsmicomb.dta

. 
. * Polity scores
. * Relationships with organizational forms 
. logit coalitionleadership newpolitymin1 newpolitymin1sq if startyear>1899, or nolog

Logistic regression                             Number of obs     =        276
                                                LR chi2(2)        =      12.50
                                                Prob > chi2       =     0.0019
Log likelihood = -144.40571                     Pseudo R2         =     0.0415

-------------------------------------------------------------------------------------
coalitionleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9455704   .0286543    -1.85   0.065     .8910442    1.003433
    newpolitymin1sq |   .9817722   .0057844    -3.12   0.002     .9705002    .9931752
              _cons |   .5223426   .1168431    -2.90   0.004     .3369378    .8097691
-------------------------------------------------------------------------------------

. *       RESULT:  Marginally significant and negative
. logit traditionalleadership newpolitymin1 newpolitymin1sq if startyear>1899, or nolog

Logistic regression                             Number of obs     =        276
                                                LR chi2(2)        =      16.13
                                                Prob > chi2       =     0.0003
Log likelihood = -66.209812                     Pseudo R2         =     0.1086

---------------------------------------------------------------------------------------
traditionalleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
        newpolitymin1 |   .8106625   .0748892    -2.27   0.023     .6764035    .9715705
      newpolitymin1sq |   .9984179   .0120561    -0.13   0.896     .9750658    1.022329
                _cons |   .0381834   .0180706    -6.90   0.000     .0151021    .0965412
---------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and negative
. * Relationship with goals 
. logit democrat newpolitymin1 newpolitymin1sq if startyear>1899, or nolog

Logistic regression                             Number of obs     =        276
                                                LR chi2(2)        =      11.98
                                                Prob > chi2       =     0.0025
Log likelihood = -164.43045                     Pseudo R2         =     0.0352

---------------------------------------------------------------------------------
       democrat | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
  newpolitymin1 |   .9219938   .0245097    -3.06   0.002     .8751858    .9713053
newpolitymin1sq |   .9908065   .0050619    -1.81   0.071     .9809348    1.000778
          _cons |   .5304662   .1139679    -2.95   0.003     .3481621    .8082281
---------------------------------------------------------------------------------

. *       RESULT:  Marginally significant and negative
. logit antimonarch newpolitymin1 newpolitymin1sq if startyear>1899, or nolog

Logistic regression                             Number of obs     =        276
                                                LR chi2(2)        =       8.14
                                                Prob > chi2       =     0.0171
Log likelihood = -79.799424                     Pseudo R2         =     0.0485

---------------------------------------------------------------------------------
    antimonarch | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
  newpolitymin1 |   .9110171   .0406753    -2.09   0.037     .8346836    .9943315
newpolitymin1sq |   1.006115   .0082062     0.75   0.455     .9901586    1.022328
          _cons |   .0594431   .0235654    -7.12   0.000     .0273309    .1292853
---------------------------------------------------------------------------------

. *       RESULT:  Statistically significant
. * Relationship with participation
. reg lnparticnum newpolitymin1 newpolitymin1sq if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        256
                                                F(2, 253)         =       1.89
                                                Prob > F          =     0.1538
                                                R-squared         =     0.0159
                                                Root MSE          =     1.8598

---------------------------------------------------------------------------------
                |               Robust
    lnparticnum |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
  newpolitymin1 |  -.0378198    .020012    -1.89   0.060    -.0772311    .0015915
newpolitymin1sq |    .001151   .0043045     0.27   0.789    -.0073261    .0096281
          _cons |   10.49112   .1886365    55.62   0.000     10.11963    10.86262
---------------------------------------------------------------------------------

. *       RESULT:  Statistically significant
. * Relationship with violence
. reg deathtile10 newpolitymin1 newpolitymin1sq if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        263
                                                F(2, 260)         =       0.59
                                                Prob > F          =     0.5549
                                                R-squared         =     0.0046
                                                Root MSE          =      2.896

---------------------------------------------------------------------------------
                |               Robust
    deathtile10 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
  newpolitymin1 |  -.0264841   .0297762    -0.89   0.375    -.0851173    .0321492
newpolitymin1sq |   .0032467   .0061489     0.53   0.598    -.0088613    .0153547
          _cons |   5.054528   .2883305    17.53   0.000     4.486768    5.622289
---------------------------------------------------------------------------------

. *       RESULT:  No relationship
. logit urbandum newpolitymin1 newpolitymin1sq if startyear>1899, or nolog

Logistic regression                             Number of obs     =        276
                                                LR chi2(2)        =       2.60
                                                Prob > chi2       =     0.2723
Log likelihood = -186.48569                     Pseudo R2         =     0.0069

---------------------------------------------------------------------------------
       urbandum | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
  newpolitymin1 |   .9677714   .0199366    -1.59   0.112     .9294748    1.007646
newpolitymin1sq |   .9979898   .0043034    -0.47   0.641     .9895909     1.00646
          _cons |   1.422953   .2882766     1.74   0.082     .9566328    2.116585
---------------------------------------------------------------------------------

. *       RESULT:  No relationship
. 
. * Years incumbent in power
. * Relationship with organizational forms 
. logit coalitionleadership newincumbpowerdur if startyear>1899, or nolog

Logistic regression                             Number of obs     =        286
                                                LR chi2(1)        =       3.51
                                                Prob > chi2       =     0.0611
Log likelihood = -153.93823                     Pseudo R2         =     0.0113

-------------------------------------------------------------------------------------
coalitionleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
  newincumbpowerdur |    1.02872   .0152783     1.91   0.057     .9992072    1.059105
              _cons |   .2430972   .0462547    -7.43   0.000     .1674248    .3529719
-------------------------------------------------------------------------------------

. *       RESULT:  Marginally significant and positive
. logit traditionalleadership newincumbpowerdur if startyear>1899, or nolog

Logistic regression                             Number of obs     =        286
                                                LR chi2(1)        =       4.47
                                                Prob > chi2       =     0.0345
Log likelihood = -72.815701                     Pseudo R2         =     0.0298

---------------------------------------------------------------------------------------
traditionalleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
    newincumbpowerdur |   1.048034   .0220368     2.23   0.026     1.005721    1.092128
                _cons |    .051216   .0168216    -9.05   0.000      .026905    .0974941
---------------------------------------------------------------------------------------

. *       RESULT:  Statistcally significant and positive
. * Relationship with goals
. logit democrat newincumbpowerdur if startyear>1899, or nolog

Logistic regression                             Number of obs     =        286
                                                LR chi2(1)        =      11.93
                                                Prob > chi2       =     0.0006
Log likelihood =  -169.7452                     Pseudo R2         =     0.0340

-----------------------------------------------------------------------------------
         democrat | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newincumbpowerdur |    1.05017   .0150154     3.42   0.001     1.021149    1.080016
            _cons |   .2935186   .0526027    -6.84   0.000       .20658    .4170451
-----------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. logit antimonarch newincumbpowerdur if startyear>1899, or nolog

Logistic regression                             Number of obs     =        286
                                                LR chi2(1)        =      22.07
                                                Prob > chi2       =     0.0000
Log likelihood = -73.765594                     Pseudo R2         =     0.1301

-----------------------------------------------------------------------------------
      antimonarch | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newincumbpowerdur |    1.09636   .0212895     4.74   0.000     1.055417    1.138891
            _cons |   .0350869   .0124622    -9.43   0.000     .0174909    .0703843
-----------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. * Relationship with participation
. reg lnparticnum newincumbpowerdur if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        265
                                                F(1, 263)         =       7.94
                                                Prob > F          =     0.0052
                                                R-squared         =     0.0332
                                                Root MSE          =     1.8511

-----------------------------------------------------------------------------------
                  |               Robust
      lnparticnum |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newincumbpowerdur |   .0390699   .0138646     2.82   0.005     .0117701    .0663697
            _cons |   10.24445   .1449892    70.66   0.000     9.958963    10.52994
-----------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. * Relationship with violence
. reg deathtile10 newincumbpowerdur if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        273
                                                F(1, 271)         =       3.22
                                                Prob > F          =     0.0737
                                                R-squared         =     0.0115
                                                Root MSE          =     2.9008

-----------------------------------------------------------------------------------
                  |               Robust
      deathtile10 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newincumbpowerdur |  -.0354329    .019733    -1.80   0.074    -.0742823    .0034165
            _cons |   5.536687    .232047    23.86   0.000     5.079843     5.99353
-----------------------------------------------------------------------------------

. *       RESULT:  Marginally significant and negative
. logit urbandum newincumbpowerdur if startyear>1899, or nolog

Logistic regression                             Number of obs     =        286
                                                LR chi2(1)        =       9.20
                                                Prob > chi2       =     0.0024
Log likelihood = -190.54447                     Pseudo R2         =     0.0236

-----------------------------------------------------------------------------------
         urbandum | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newincumbpowerdur |    1.04542   .0161623     2.87   0.004     1.014218    1.077583
            _cons |   .9769938   .1560047    -0.15   0.884     .7144525    1.336012
-----------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. 
. * GDP per capita
. * Relationship with organizational forms
. logit coalitionleadership newgdppcthl if startyear>1899, or nolog

Logistic regression                             Number of obs     =        283
                                                LR chi2(1)        =      16.61
                                                Prob > chi2       =     0.0000
Log likelihood = -146.58406                     Pseudo R2         =     0.0536

-------------------------------------------------------------------------------------
coalitionleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        newgdppcthl |   1.228382   .0622773     4.06   0.000      1.11219    1.356714
              _cons |    .165415   .0366242    -8.13   0.000     .1071795    .2552923
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. logit traditionalleadership newgdppcthl if startyear>1899, or nolog

Logistic regression                             Number of obs     =        283
                                                LR chi2(1)        =       2.85
                                                Prob > chi2       =     0.0916
Log likelihood = -68.243956                     Pseudo R2         =     0.0204

---------------------------------------------------------------------------------------
traditionalleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
          newgdppcthl |   .8230726   .1085911    -1.48   0.140     .6355298    1.065959
                _cons |   .1119893    .038715    -6.33   0.000      .056874    .2205155
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant and negative
. * Relationship with goals 
. logit democrat newgdppcthl if startyear>1899, or nolog

Logistic regression                             Number of obs     =        283
                                                LR chi2(1)        =      17.09
                                                Prob > chi2       =     0.0000
Log likelihood = -164.41649                     Pseudo R2         =     0.0494

------------------------------------------------------------------------------
    democrat | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 newgdppcthl |   1.221781   .0603426     4.06   0.000     1.109056    1.345964
       _cons |   .2381915   .0481736    -7.09   0.000     .1602408    .3540622
------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. logit antimonarch newgdppcthl if startyear>1899, or nolog

Logistic regression                             Number of obs     =        283
                                                LR chi2(1)        =       3.30
                                                Prob > chi2       =     0.0694
Log likelihood = -82.877857                     Pseudo R2         =     0.0195

------------------------------------------------------------------------------
 antimonarch | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 newgdppcthl |   .8349321     .09355    -1.61   0.107     .6703139    1.039978
       _cons |   .1470897   .0447759    -6.30   0.000     .0809965     .267115
------------------------------------------------------------------------------

. *       RESULT:  Not sigificant and negative
. * Relationship with participation
. reg lnparticnum newgdppcthl if startyear>1899, vce(robust) 

Linear regression                               Number of obs     =        262
                                                F(1, 260)         =      31.02
                                                Prob > F          =     0.0000
                                                R-squared         =     0.1129
                                                Root MSE          =     1.7843

------------------------------------------------------------------------------
             |               Robust
 lnparticnum |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 newgdppcthl |   .2409025   .0432536     5.57   0.000     .1557306    .3260744
       _cons |   9.896178   .1572558    62.93   0.000     9.586521    10.20584
------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. * Relationship with violence 
. reg deathtile10 newgdppcthl if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        271
                                                F(1, 269)         =      78.64
                                                Prob > F          =     0.0000
                                                R-squared         =     0.1561
                                                Root MSE          =     2.6942

------------------------------------------------------------------------------
             |               Robust
 deathtile10 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 newgdppcthl |  -.4408263   .0497093    -8.87   0.000    -.5386951   -.3429575
       _cons |   6.490542   .2368805    27.40   0.000     6.024167    6.956918
------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and negative
. logit urbandum newgdppcthl if startyear>1899, or nolog

Logistic regression                             Number of obs     =        283
                                                LR chi2(1)        =      52.08
                                                Prob > chi2       =     0.0000
Log likelihood = -166.52804                     Pseudo R2         =     0.1352

------------------------------------------------------------------------------
    urbandum | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 newgdppcthl |   1.585077   .1297912     5.63   0.000     1.350054    1.861013
       _cons |   .4746003   .0966713    -3.66   0.000     .3183808    .7074718
------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. 
. * Oil production
. * Relationship with organizational forms
. logit coalitionleadership newlnoill if startyear>1899, or nolog

Logistic regression                             Number of obs     =        285
                                                LR chi2(1)        =       0.64
                                                Prob > chi2       =     0.4231
Log likelihood =  -155.1042                     Pseudo R2         =     0.0021

-------------------------------------------------------------------------------------
coalitionleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
          newlnoill |   .9753084   .0306681    -0.80   0.427     .9170148    1.037308
              _cons |   .3373192   .0605964    -6.05   0.000     .2372085    .4796802
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant and negative
. logit traditionalleadership newlnoill if startyear>1899, or nolog

Logistic regression                             Number of obs     =        285
                                                LR chi2(1)        =       1.75
                                                Prob > chi2       =     0.1854
Log likelihood =  -71.53938                     Pseudo R2         =     0.0121

---------------------------------------------------------------------------------------
traditionalleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
            newlnoill |   1.067695   .0523712     1.34   0.182     .9698287    1.175436
                _cons |   .0563625   .0191904    -8.45   0.000     .0289182     .109852
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant and positive
. * Relationship with goals 
. logit democrat newlnoill if startyear>1899, or nolog

Logistic regression                             Number of obs     =        285
                                                LR chi2(1)        =       1.82
                                                Prob > chi2       =     0.1773
Log likelihood = -173.60785                     Pseudo R2         =     0.0052

------------------------------------------------------------------------------
    democrat | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   newlnoill |   .9616986   .0281367    -1.33   0.182      .908103    1.018457
       _cons |   .4992831   .0826936    -4.19   0.000     .3608834    .6907595
------------------------------------------------------------------------------

. *       RESULT:  Not significant and negative
. logit antimonarch newlnoill if startyear>1899, or nolog

Logistic regression                             Number of obs     =        285
                                                LR chi2(1)        =       0.18
                                                Prob > chi2       =     0.6722
Log likelihood = -82.256798                     Pseudo R2         =     0.0011

------------------------------------------------------------------------------
 antimonarch | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   newlnoill |   .9799835   .0472488    -0.42   0.675     .8916184    1.077106
       _cons |   .0990743   .0270107    -8.48   0.000     .0580627    .1690538
------------------------------------------------------------------------------

. *       RESULT:  Not significant and negative
. * Relationship with participation
. reg lnparticnum newlnoill if startyear>1899, vce(robust) 

Linear regression                               Number of obs     =        264
                                                F(1, 262)         =       5.37
                                                Prob > F          =     0.0213
                                                R-squared         =     0.0201
                                                Root MSE          =     1.8586

------------------------------------------------------------------------------
             |               Robust
 lnparticnum |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   newlnoill |   .0580327   .0250491     2.32   0.021     .0087095    .1073558
       _cons |   10.32819   .1364393    75.70   0.000     10.05954    10.59685
------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. * Relationship with violence 
. reg deathtile10 newlnoill if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        272
                                                F(1, 270)         =       0.07
                                                Prob > F          =     0.7869
                                                R-squared         =     0.0002
                                                Root MSE          =     2.9284

------------------------------------------------------------------------------
             |               Robust
 deathtile10 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   newlnoill |   .0099893   .0369058     0.27   0.787    -.0626704     .082649
       _cons |   5.203722   .2416652    21.53   0.000     4.727934    5.679509
------------------------------------------------------------------------------

. *       RESULT:  Not significant and positive
. * Relationship with urban location of contention
. logit urbandum newlnoill if startyear>1899, or nolog

Logistic regression                             Number of obs     =        285
                                                LR chi2(1)        =       2.33
                                                Prob > chi2       =     0.1268
Log likelihood = -193.12484                     Pseudo R2         =     0.0060

------------------------------------------------------------------------------
    urbandum | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   newlnoill |   1.041554     .02795     1.52   0.129     .9881889    1.097801
       _cons |   1.161594   .1814686     0.96   0.338     .8552174    1.577729
------------------------------------------------------------------------------

. *       RESULT:  Not significant and positive
. 
. * Military expenditures per soldier (deciles)
. * Years incumbent in power
. * Relationship with organizational forms
. logit coalitionleadership newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        240
                                                LR chi2(1)        =      15.35
                                                Prob > chi2       =     0.0001
Log likelihood = -127.28432                     Pseudo R2         =     0.0569

-------------------------------------------------------------------------------------
coalitionleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
newmilexpsold10tile |   1.236081   .0702093     3.73   0.000     1.105857    1.381641
              _cons |   .0900594   .0364953    -5.94   0.000     .0406997    .1992811
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. logit traditionalleadership newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        240
                                                LR chi2(1)        =       1.89
                                                Prob > chi2       =     0.1686
Log likelihood = -55.162544                     Pseudo R2         =     0.0169

---------------------------------------------------------------------------------------
traditionalleadership | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
  newmilexpsold10tile |   .8804516    .082623    -1.36   0.175     .7325332    1.058239
                _cons |   .1302989   .0682808    -3.89   0.000     .0466535     .363913
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant and negative
. * Relationship with goals 
. logit democrat newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        240
                                                LR chi2(1)        =      14.45
                                                Prob > chi2       =     0.0001
Log likelihood = -137.64687                     Pseudo R2         =     0.0499

-------------------------------------------------------------------------------------
           democrat | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
newmilexpsold10tile |   1.213959   .0644229     3.65   0.000     1.094038    1.347026
              _cons |   .1268391    .046946    -5.58   0.000     .0614052    .2620003
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. logit antimonarch newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        240
                                                LR chi2(1)        =       9.42
                                                Prob > chi2       =     0.0021
Log likelihood = -64.130941                     Pseudo R2         =     0.0684

-------------------------------------------------------------------------------------
        antimonarch | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
newmilexpsold10tile |   .7704435   .0700002    -2.87   0.004     .6447676    .9206158
              _cons |   .3213019   .1400042    -2.61   0.009     .1367762    .7547724
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and negative
. * Relationship with participation
. reg lnparticnum newmilexpsold10tile if startyear>1899, vce(robust) 

Linear regression                               Number of obs     =        227
                                                F(1, 225)         =      14.12
                                                Prob > F          =     0.0002
                                                R-squared         =     0.0678
                                                Root MSE          =     1.8767

-------------------------------------------------------------------------------------
                    |               Robust
        lnparticnum |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
newmilexpsold10tile |   .1739567   .0463018     3.76   0.000      .082716    .2651974
              _cons |   9.602057   .3040863    31.58   0.000     9.002835    10.20128
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. * Relationship with violence 
. reg deathtile10 newmilexpsold10tile if startyear>1899, vce(robust)

Linear regression                               Number of obs     =        230
                                                F(1, 228)         =      19.52
                                                Prob > F          =     0.0000
                                                R-squared         =     0.0712
                                                Root MSE          =     2.8344

-------------------------------------------------------------------------------------
                    |               Robust
        deathtile10 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
newmilexpsold10tile |  -.2678575    .060619    -4.42   0.000    -.3873025   -.1484124
              _cons |   6.764598   .3893985    17.37   0.000     5.997318    7.531878
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and negative
. * Relationship with urban location of contention
. logit urbandum newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        240
                                                LR chi2(1)        =      11.52
                                                Prob > chi2       =     0.0007
Log likelihood = -157.57329                     Pseudo R2         =     0.0353

-------------------------------------------------------------------------------------
           urbandum | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
newmilexpsold10tile |   1.168329   .0546668     3.32   0.001      1.06595     1.28054
              _cons |   .5727859   .1671786    -1.91   0.056     .3232606     1.01492
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant and positive
. 
. 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++
. * ROBUSTNESS TESTS FOR COMBINED MODEL (2)in Table 4.3
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++
. 
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Bootstrapped standard errors--complete case sample--combined model (2)
. * ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. use revolutionaryepsmicomb.dta, clear

. eststo: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpo
> werdur newgdppcthl newlnoill newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        212
                                                LR chi2(10)       =      96.80
                                                Prob > chi2       =     0.0000
Log likelihood = -93.516594                     Pseudo R2         =     0.3410

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |    1.62341   .2276605     3.46   0.001     1.233272    2.136964
           urbandum |   56.47999   87.38278     2.61   0.009     2.722461     1171.73
        deathtile10 |   1.284022   .2162003     1.48   0.138     .9231011    1.786058
    urbxdeathtile10 |   .5270382   .1078415    -3.13   0.002     .3529157    .7870698
      newpolitymin1 |   .9141814   .0344588    -2.38   0.017     .8490779    .9842768
    newpolitymin1sq |   .9773756   .0077007    -2.90   0.004     .9623986    .9925858
  newincumbpowerdur |   1.059527    .024634     2.49   0.013     1.012328    1.108925
        newgdppcthl |   .7414142   .0730933    -3.03   0.002     .6111447    .8994515
          newlnoill |   .8828805   .0398304    -2.76   0.006     .8081662    .9645019
newmilexpsold10tile |   1.288978   .1046416     3.13   0.002     1.099369    1.511289
              _cons |   .0003419   .0005859    -4.66   0.000     .0000119    .0098311
-------------------------------------------------------------------------------------
(est1 stored)

. eststo: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpo
> werdur newgdppcthl newlnoill newmilexpsold10tile if startyear>1899, or nolog vce(bootstrap, bca seed(1234) rep(1
> 000))
(running logit on estimation sample)

Jackknife replications (212)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200
............

Bootstrap replications (1000)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100
..................................................   150
..................................................   200
..................................................   250
..................................................   300
..................................................   350
..................................................   400
..................................................   450
..................................................   500
..................................................   550
..................................................   600
..................................................   650
..................................................   700
..................................................   750
..................................................   800
..................................................   850
..................................................   900
..................................................   950
..................................................  1000

Logistic regression                             Number of obs     =        212
                                                Replications      =      1,000
                                                Wald chi2(10)     =      37.59
                                                Prob > chi2       =     0.0000
Log likelihood = -93.516594                     Pseudo R2         =     0.3410

-------------------------------------------------------------------------------------
                    |   Observed   Bootstrap                         Normal-based
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |    1.62341   .2871304     2.74   0.006     1.147829    2.296038
           urbandum |   56.47999   116.5077     1.96   0.051     .9908881    3219.323
        deathtile10 |   1.284022   .2820324     1.14   0.255      .834849    1.974862
    urbxdeathtile10 |   .5270382   .1381788    -2.44   0.015     .3152639    .8810693
      newpolitymin1 |   .9141814   .0378463    -2.17   0.030     .8429338    .9914512
    newpolitymin1sq |   .9773756    .008191    -2.73   0.006     .9614526    .9935623
  newincumbpowerdur |   1.059527   .0251887     2.43   0.015      1.01129    1.110064
        newgdppcthl |   .7414142   .0838786    -2.64   0.008     .5939662    .9254652
          newlnoill |   .8828805    .041935    -2.62   0.009     .8043991    .9690188
newmilexpsold10tile |   1.288978   .1153847     2.84   0.005     1.081556    1.536179
              _cons |   .0003419   .0007749    -3.52   0.000     4.02e-06    .0290473
-------------------------------------------------------------------------------------
(est2 stored)

. estat bootstrap, all

Logistic regression                             Number of obs     =        212
                                                Replications      =       1000

------------------------------------------------------------------------------
             |    Observed               Bootstrap
     success |       Coef.       Bias    Std. Err.  [95% Conf. Interval]
-------------+----------------------------------------------------------------
 lnparticnum |   .48452868   .0487172   .17686874    .1378723    .831185   (N)
             |                                       .2203484   .8972531   (P)
             |                                       .1696774   .8049699  (BC)
             |                                       .1435743   .7820542 (BCa)
    urbandum |   4.0338864   .4156371   2.0628135   -.0091537   8.076926   (N)
             |                                       .9108429   9.041993   (P)
             |                                       .4834161   8.024796  (BC)
             |                                      -.5106276    7.55625 (BCa)
 deathtile10 |   .24999715   .0233029   .21964767   -.1805044   .6804987   (N)
             |                                      -.1045126   .7462111   (P)
             |                                      -.1102389   .7406856  (BC)
             |                                      -.1519583   .6646836 (BCa)
urbxdeath~10 |  -.64048217  -.0677571   .26217991   -1.154345   -.126619   (N)
             |                                      -1.301646   -.252541   (P)
             |                                      -1.128582  -.1521667  (BC)
             |                                      -1.098839   -.138392 (BCa)
newpolitym~1 |  -.08972622  -.0078015   .04139907   -.1708669  -.0085855   (N)
             |                                      -.1831534   -.018934   (P)
             |                                      -.1643021  -.0040615  (BC)
             |                                      -.1617505  -.0032693 (BCa)
newpolitym~q |  -.02288422  -.0020183   .00838063     -.03931  -.0064585   (N)
             |                                      -.0418709  -.0100317   (P)
             |                                      -.0382797  -.0074299  (BC)
             |                                      -.0377763  -.0073397 (BCa)
newincumbp~r |   .05782228   .0069039    .0237735    .0112271   .1044175   (N)
             |                                       .0218938   .1136142   (P)
             |                                       .0155239   .1008603  (BC)
             |                                       .0130564   .0998999 (BCa)
 newgdppcthl |   -.2991958  -.0300308   .11313323   -.5209329  -.0774587   (N)
             |                                      -.5685908    -.12153   (P)
             |                                      -.4915231  -.0689456  (BC)
             |                                      -.4858027  -.0552521 (BCa)
   newlnoill |  -.12456548  -.0105138   .04749794   -.2176597  -.0314712   (N)
             |                                      -.2312456  -.0432062   (P)
             |                                      -.2089972   -.021639  (BC)
             |                                      -.2079752  -.0199533 (BCa)
newmilexps~e |   .25384949   .0227076   .08951641    .0784006   .4292984   (N)
             |                                       .1103874   .4703708   (P)
             |                                       .0729632   .4152942  (BC)
             |                                       .0612431   .4132622 (BCa)
       _cons |  -7.9809385  -.8086192   2.2664242   -12.42305  -3.538829   (N)
             |                                      -13.89697  -4.745232   (P)
             |                                      -11.76163  -3.826504  (BC)
             |                                      -11.54329  -3.518927 (BCa)
------------------------------------------------------------------------------
(N)    normal confidence interval
(P)    percentile confidence interval
(BC)   bias-corrected confidence interval
(BCa)  bias-corrected and accelerated confidence interval

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(Orig Boot)

--------------------------------------------
                      (1)             (2)   
                     Orig            Boot   
--------------------------------------------
success                                     
lnparticnum         0.485***        0.485** 
                   (3.46)          (2.74)   

urbandum            4.034**         4.034+  
                   (2.61)          (1.96)   

deathtile10         0.250           0.250   
                   (1.48)          (1.14)   

urbxdeath~10       -0.640**        -0.640*  
                  (-3.13)         (-2.44)   

newpolitym~1      -0.0897*        -0.0897*  
                  (-2.38)         (-2.17)   

newpolitym~q      -0.0229**       -0.0229** 
                  (-2.90)         (-2.73)   

newincumbp~r       0.0578*         0.0578*  
                   (2.49)          (2.43)   

newgdppcthl        -0.299**        -0.299** 
                  (-3.03)         (-2.64)   

newlnoill          -0.125**        -0.125** 
                  (-2.76)         (-2.62)   

newmilexps~e        0.254**         0.254** 
                   (3.13)          (2.84)   

_cons              -7.981***       -7.981***
                  (-4.66)         (-3.52)   
--------------------------------------------
N                     212             212   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  no changes in signs or patterns of statistical significance
. eststo clear

. 
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * Bootstrapping multiple imputation estimation--combined model (2)
. * +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
. * using 10 imputations, 100 bootstraps (30 imputations takes a very long time)
. * WARNING:  This test can take up to 50 minutes to complete execution
. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. program define myboot, rclass
  1.    mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) 
> newlnoill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmile
> xp (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 = success civilw
> ar urbandum democrat antimonarch, add(10) rseed(1234) force
  2.    mi estimate, eform: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolity
> min1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile, or
  3.    return scalar b_lnp = el(e(b_mi),1,1)
  4.    return scalar b_urb = el(e(b_mi),1,2)
  5.    return scalar b_d10 = el(e(b_mi),1,3)
  6.    return scalar b_uxd = el(e(b_mi),1,4)
  7.    return scalar b_npo1 = el(e(b_mi),1,5)
  8.    return scalar b_npsq = el(e(b_mi),1,6)
  9.    return scalar b_newi = el(e(b_mi),1,7)
 10.    return scalar b_newg = el(e(b_mi),1,8)
 11.    return scalar b_newo = el(e(b_mi),1,9)
 12.    return scalar b_newm = el(e(b_mi),1,10)
 13.    return scalar b_int  = el(e(b_mi),1,11)
 14. end

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum urbandum deathtile10 urbxdeathtile10 newcivxmilexp newpolitymin1 newpolitymin1sq
>  newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile

. set seed 1234

. bootstrap b_lnparticnum=r(b_lnp) b_urbandum=r(b_urb) b_deathtile10=r(b_d10) b_urbxdeathtile10=r(b_uxd) b_newpoli
> tymin1=r(b_npo1) b_newpolitymin1sq=r(b_npsq) b_newincumbpowerdur=r(b_newi) b_newgdppcthl=r(b_newg) b_newlnoill=r
> (b_newo) b_newmilexpsold10tile=r(b_newm) intercept=r(b_int), reps(100): myboot 
(running myboot on estimation sample)

Warning:  Because myboot is not an estimation command or does not set e(sample), bootstrap has no way to
          determine which observations are used in calculating the statistics and so assumes that all
          observations are used.  This means that no observations will be excluded from the resampling because
          of missing values or other reasons.

          If the assumption is not true, press Break, save the data, and drop the observations that are to be
          excluded.  Be sure that the dataset in memory contains only the relevant data.

Bootstrap replications (100)
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..................................................    50
..................................................   100

Bootstrap results                               Number of obs     =        288
                                                Replications      =        100

      command:  myboot
b_lnparticnum:  r(b_lnp)
   b_urbandum:  r(b_urb)
b_deathtile10:  r(b_d10)
b_urbxdeat~10:  r(b_uxd)
b_newpolity~1:  r(b_npo1)
b_newpolity~q:  r(b_npsq)
b_newincumb~r:  r(b_newi)
b_newgdppcthl:  r(b_newg)
  b_newlnoill:  r(b_newo)
b_newmilexp~e:  r(b_newm)
    intercept:  r(b_int)

---------------------------------------------------------------------------------------
                      |   Observed   Bootstrap                         Normal-based
                      |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
        b_lnparticnum |   .5391207   .1308481     4.12   0.000     .2826632    .7955783
           b_urbandum |   3.552408   1.705524     2.08   0.037     .2096422    6.895174
        b_deathtile10 |   .2237078   .1906605     1.17   0.241    -.1499798    .5973955
    b_urbxdeathtile10 |  -.5650885   .1920039    -2.94   0.003    -.9414092   -.1887678
      b_newpolitymin1 |  -.1101031   .0320994    -3.43   0.001    -.1730168   -.0471894
    b_newpolitymin1sq |  -.0212164   .0065336    -3.25   0.001     -.034022   -.0084107
  b_newincumbpowerdur |   .0375243   .0168867     2.22   0.026      .004427    .0706216
        b_newgdppcthl |  -.2720538    .099334    -2.74   0.006    -.4667449   -.0773627
          b_newlnoill |  -.1113781   .0397215    -2.80   0.005    -.1892309   -.0335253
b_newmilexpsold10tile |   .2478312   .0701396     3.53   0.000       .11036    .3853024
            intercept |  -8.401892   1.761237    -4.77   0.000    -11.85385   -4.949931
---------------------------------------------------------------------------------------

. * Displaying exponentiated form
. bootstrap, eform

Bootstrap results                               Number of obs     =        288
                                                Replications      =        100

      command:  myboot
b_lnparticnum:  r(b_lnp)
   b_urbandum:  r(b_urb)
b_deathtile10:  r(b_d10)
b_urbxdeat~10:  r(b_uxd)
b_newpolity~1:  r(b_npo1)
b_newpolity~q:  r(b_npsq)
b_newincumb~r:  r(b_newi)
b_newgdppcthl:  r(b_newg)
  b_newlnoill:  r(b_newo)
b_newmilexp~e:  r(b_newm)
    intercept:  r(b_int)

---------------------------------------------------------------------------------------
                      |   Observed   Bootstrap                         Normal-based
                      |     exp(b)   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
        b_lnparticnum |   1.714499   .2243389     4.12   0.000     1.326658    2.215722
           b_urbandum |   34.89726   59.51812     2.08   0.037     1.233237    987.4979
        b_deathtile10 |   1.250706   .2384601     1.17   0.241     .8607254    1.817379
    b_urbxdeathtile10 |   .5683099   .1091177    -2.94   0.003     .3900777    .8279788
      b_newpolitymin1 |   .8957418   .0287528    -3.43   0.001     .8411235    .9539067
    b_newpolitymin1sq |   .9790071   .0063965    -3.25   0.001     .9665502    .9916246
  b_newincumbpowerdur |   1.038237   .0175324     2.22   0.026     1.004437    1.073175
        b_newgdppcthl |   .7618133    .075674    -2.74   0.006       .62704    .9255541
          b_newlnoill |   .8946004   .0355349    -2.80   0.005     .8275954    .9670304
b_newmilexpsold10tile |   1.281244    .089866     3.53   0.000      1.11668    1.470059
            intercept |   .0002244   .0003953    -4.77   0.000     7.11e-06    .0070839
---------------------------------------------------------------------------------------

. clear programs

. * Result: All variables statistically significant, with exception of deathtile10
. 
. * ++++++++++++++++++++++++++++++++++++++++
. * Visual inspection of potential outliers
. * ++++++++++++++++++++++++++++++++++++++++
. 
. * Complete case sample
. use revolutionaryepsmicomb.dta, clear

. *  Create predictions
. quietly: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbp
> owerdur newgdppcthl newlnoill newmilexpsold10tile if startyear>1899, or nolog

. predict pr , pr
(76 missing values generated)

. predict stdres, rstand
(76 missing values generated)

. predict dev, dev
(76 missing values generated)

. predict hat, hat
(76 missing values generated)

. predict dx2, dx2
(76 missing values generated)

. predict dd, dd
(76 missing values generated)

. * Standardized Pearson residuals by predicted probability
. scatter stdres pr, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat1.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat1.pdf written in PDF format)

. *       The following are potential outliers:
. *               revid 269 (1990 Mongolian Revolution), revid 106 (Solidarity Uprising), revid 195 (Togo 1991 Rev
> olution)
. * Standardized Pearson residuals by revid
. scatter stdres revid, mlab(revi)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat2.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat2.pdf written in PDF format)

. *       The following are potential outliers:
. *               revid 35 (Albanian Uprising of 1924) and revid 106 (Solidarity Uprising)
. * Deviance residual by revid
. scatter dev revid, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat3.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat3.pdf written in PDF format)

. *       The following are potential outliers:
. *               revid 35 (Albanian Uprising of 1924) and revid 106 (Solidarity Uprising)
. * Leverage by predicted probability
. scatter hat pr, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat4.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat4.pdf written in PDF format)

. *       The following are potential outliers:
. *               revid 372 (Second Tuareg Rebellion in Mali) and revid 102 (Iranian Revolution)
. * Leverage by revid (with cutoff point of 3 * the mean of leverage)
. mean hat

Mean estimation                   Number of obs   =        212

--------------------------------------------------------------
             |       Mean   Std. Err.     [95% Conf. Interval]
-------------+------------------------------------------------
         hat |   .0518868   .0021241      .0476996     .056074
--------------------------------------------------------------

. matrix coefs = e(b)

. local hatmean = 3 * (coefs[1,1])

. scatter hat revid, mlab(revid) yline(0) yline(`hatmean')

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat5.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat5.pdf written in PDF format)

. *       The following are potential outliers:
. *               revid 372 (Second Tuareg Rebellion in Mali) and revid 102 (Iranian Revolution)
. * Difference of chi-squares
. scatter dx2 revid, mlab(revid)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat6.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat6.pdf written in PDF format)

. *       The following are potential outliers:
. *       revid 106 (Solidarity Uprising) and revid 35 (Albanian Uprising of 1924)
. * Difference of deviances
. scatter dd revid, mlab(revid)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat7.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat7.pdf written in PDF format)

. *       The following are potential outliers:
. *       revid 106 (Solidarity Uprising) and revid 35 (Albanian Uprising of 1924)
. * Drop predictions
. drop pr stdres dev hat dx2 dd

. * Checking to see if dropping potential outliers alters any findings 
. eststo: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpo
> werdur newgdppcthl newlnoill newmilexpsold10tile if startyear>1899, or nolog

Logistic regression                             Number of obs     =        212
                                                LR chi2(10)       =      96.80
                                                Prob > chi2       =     0.0000
Log likelihood = -93.516594                     Pseudo R2         =     0.3410

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |    1.62341   .2276605     3.46   0.001     1.233272    2.136964
           urbandum |   56.47999   87.38278     2.61   0.009     2.722461     1171.73
        deathtile10 |   1.284022   .2162003     1.48   0.138     .9231011    1.786058
    urbxdeathtile10 |   .5270382   .1078415    -3.13   0.002     .3529157    .7870698
      newpolitymin1 |   .9141814   .0344588    -2.38   0.017     .8490779    .9842768
    newpolitymin1sq |   .9773756   .0077007    -2.90   0.004     .9623986    .9925858
  newincumbpowerdur |   1.059527    .024634     2.49   0.013     1.012328    1.108925
        newgdppcthl |   .7414142   .0730933    -3.03   0.002     .6111447    .8994515
          newlnoill |   .8828805   .0398304    -2.76   0.006     .8081662    .9645019
newmilexpsold10tile |   1.288978   .1046416     3.13   0.002     1.099369    1.511289
              _cons |   .0003419   .0005859    -4.66   0.000     .0000119    .0098311
-------------------------------------------------------------------------------------
(est1 stored)

. eststo: logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpo
> werdur newgdppcthl newlnoill newmilexpsold10tile if startyear>1899 & revid!=35 & revid~=106 & revid~=372 & revid
> ~=102 & revid~=269 & revid~=195, or nolog

Logistic regression                             Number of obs     =        206
                                                LR chi2(10)       =     113.43
                                                Prob > chi2       =     0.0000
Log likelihood = -81.335412                     Pseudo R2         =     0.4108

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.713983   .2621542     3.52   0.000     1.270033    2.313119
           urbandum |   214.1264   392.5196     2.93   0.003     5.892702    7780.828
        deathtile10 |   1.400862   .2777931     1.70   0.089      .949731    2.066283
    urbxdeathtile10 |   .4487501   .1059071    -3.40   0.001     .2825641    .7126761
      newpolitymin1 |   .8925072   .0365896    -2.77   0.006     .8235985    .9671813
    newpolitymin1sq |   .9774824   .0082399    -2.70   0.007     .9614652    .9937664
  newincumbpowerdur |   1.066974   .0281701     2.46   0.014     1.013166     1.12364
        newgdppcthl |   .6616967   .0740926    -3.69   0.000     .5313085    .8240834
          newlnoill |   .8563917   .0423534    -3.13   0.002     .7772768    .9435593
newmilexpsold10tile |   1.445281   .1333175     3.99   0.000     1.206242    1.731689
              _cons |   .0000577    .000119    -4.73   0.000     1.01e-06    .0032887
-------------------------------------------------------------------------------------
(est2 stored)

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(All_n No_outl)

--------------------------------------------
                      (1)             (2)   
                    All_n         No_outl   
--------------------------------------------
success                                     
lnparticnum         0.485***        0.539***
                   (3.46)          (3.52)   

urbandum            4.034**         5.367** 
                   (2.61)          (2.93)   

deathtile10         0.250           0.337+  
                   (1.48)          (1.70)   

urbxdeath~10       -0.640**        -0.801***
                  (-3.13)         (-3.40)   

newpolitym~1      -0.0897*         -0.114** 
                  (-2.38)         (-2.77)   

newpolitym~q      -0.0229**       -0.0228** 
                  (-2.90)         (-2.70)   

newincumbp~r       0.0578*         0.0648*  
                   (2.49)          (2.46)   

newgdppcthl        -0.299**        -0.413***
                  (-3.03)         (-3.69)   

newlnoill          -0.125**        -0.155** 
                  (-2.76)         (-3.13)   

newmilexps~e        0.254**         0.368***
                   (3.13)          (3.99)   

_cons              -7.981***       -9.760***
                  (-4.66)         (-4.73)   
--------------------------------------------
N                     212             206   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  deathtile10 grows marginally significant at the .10 level; otherwise, nothing changes
. eststo clear

. 
. * Multiple imputation sample
. use revolutionaryepsmicomb.dta, clear

. quietly: mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urb
> xdeathtile10 newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile if starty
> ear>1899

. mi predict xb using miest , xb

. mi predict stdp using miest, stdp

. scatter  stdp xb, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat8.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat8.pdf written in PDF format)

. *       RESULT: revid 301 (228 Uprising) as potential outlier
. *                       revid 249 (El Salvador Revolution of 1944) as potential outlier
. *                       revid 184 (Haitian Revolution of 1946) as potential outlier
. *                       revid 51 (Guatemalan Revolution) as potential outlier
. *                       revid 6 (Young Turk Revolution) as potential outlier
. scatter  stdp revid, mlab(revid)  yline(0)

. graph export Robustnesstestfiles\Logfiles\robch4tab4_3_scat9.pdf, replace
(file Robustnesstestfiles\Logfiles\robch4tab4_3_scat9.pdf written in PDF format)

. *       RESULT: revid 301 (228 Uprising) as potential outlier
. *                       revid 249 (El Salvador Revolution of 1944) as potential outlier
. *                       revid 184 (Haitian Revolution of 1946) as potential outlier
. *                       revid 51 (Guatemalan Revolution) as potential outlier
. *                       revid 6 (Young Turk Revolution) as potential outlier
. *                       revid 407 (Sandino Rebellion) as potential outlier
. *                       reivd 403 (South Sudanese Civil War) as potential outlier
. * Drop predictions
. drop xb stdp

. * Testing the effect of dropping potential outliers on results in Model 6
. eststo: quietly: mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathti
> le10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile i
> f startyear>1899 , or 
(est1 stored)

. eststo: quietly: mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathti
> le10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile i
> f startyear>1899 & revid~=301 & revid~=249 & revid~=184 & revid~=51 & revid~=6 & revid~=407 & revid~=401, or
(est2 stored)

. esttab , star (+ 0.10 * 0.05 ** 0.01 *** 0.001) mtitles(All_n No_outl) 

--------------------------------------------
                      (1)             (2)   
                    All_n         No_outl   
--------------------------------------------
success                                     
lnparticnum         0.535***        0.552***
                   (4.24)          (4.40)   

urbandum            3.589**         3.403*  
                   (2.71)          (2.57)   

deathtile10         0.231           0.227   
                   (1.54)          (1.51)   

urbxdeath~10       -0.569***       -0.539** 
                  (-3.30)         (-3.14)   

newpolitym~1       -0.114**        -0.110** 
                  (-3.28)         (-3.12)   

newpolitym~q      -0.0212**       -0.0217** 
                  (-3.09)         (-3.11)   

newincumbp~r       0.0356+         0.0365+  
                   (1.83)          (1.86)   

newgdppcthl        -0.273**        -0.257** 
                  (-3.12)         (-2.93)   

newlnoill          -0.115**        -0.116** 
                  (-2.78)         (-2.83)   

newmilexps~e        0.263***        0.260***
                   (3.48)          (3.45)   

_cons              -8.480***       -8.591***
                  (-5.46)         (-5.54)   
--------------------------------------------
N                     288             281   
--------------------------------------------
t statistics in parentheses
+ p<0.10, * p<0.05, ** p<0.01, *** p<0.001

. *       RESULT:  No changes in signs and patterns of significance
. eststo clear

. 
. * ===========================================================================
. * FURTHER TESTS FOR OMITTED VARIABLE BIAS 
. *       using multiple imputation and complete-case samples:  
. *               pop size, econ growth, youth bulges, rough terrain (for rural 
. *               episodes only), yrs schooling, soldiers per 1000 pop, civil society 
. *               index, private ownership of economy, state capacity
. * ===========================================================================
. 
. ***********************
. * Population size, t-1
. ************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newincumbage newgdppcthl newlnoill newmilexp
> sold10tile newcivxmilexp lnpop

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        192       66.67       66.67
          1 |         96       33.33      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnpop = success civilwar, add(40) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq lnpop
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq lnpop
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq lnpop
                     newmilexpsold10tile newcivxmilexp success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq lnpop
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 lnpop
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
             lnpop: pmm lnpop newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq lnpop newcivxmilexp success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     lnpop newmilexpsold10tile success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=40 .........10.........20.........30.........40 done

Multivariate imputation                     Imputations =       40
Chained equations                                 added =       40
Imputed: m=1 through m=40                       updated =        0

Initialization: monotone                     Iterations =      400
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
             lnpop: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
             lnpop |        243           45        45 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success lnpop if startyear>1899

Imputations (40):
  .........10.........20.........30.........40 done

Multiple-imputation estimates                   Imputations       =         40
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0591
                                                Largest FMI       =     0.1060
DF adjustment:   Large sample                   DF:     min       =   3,503.00
                                                        avg       =   3,560.50
                                                        max       =   3,618.00
Model F test:       Equal FMI                   F(   1, 3503.0)   =       0.12
Within VCE type:          OIM                   Prob > F          =     0.7290

------------------------------------------------------------------------------
     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       lnpop |   .9709759   .0825401    -0.35   0.729     .8219115    1.147075
       _cons |   .8022786   .6441573    -0.27   0.784     .1662105    3.872504
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.lnpop if startyear>1899

Imputations (40):
  .........10.........20.........30.........40 done

Multiple-imputation estimates                   Imputations       =         40
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0450
                                                Largest FMI       =     0.1156
DF adjustment:   Large sample                   DF:     min       =   2,948.65
                                                        avg       =   4,319.01
                                                        max       =   5,795.37
Model F test:       Equal FMI                   F(   3,34207.0)   =       5.11
Within VCE type:          OIM                   Prob > F          =     0.0016

----------------------------------------------------------------------------------
         success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
        urbandum |
            yes  |   2.626533   4.418156     0.57   0.566     .0971084    71.04098
           lnpop |   .9457437   .1327074    -0.40   0.691     .7182625    1.245271
                 |
urbandum#c.lnpop |
            yes  |     1.0097   .1808693     0.05   0.957     .7106924    1.434509
                 |
           _cons |   .5400863   .7057194    -0.47   0.637     .0416652    7.000882
----------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp lnpop if startyear>1899

Imputations (40):
  .........10.........20.........30.........40 done

Multiple-imputation estimates                   Imputations       =         40
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0908
                                                Largest FMI       =     0.1729
DF adjustment:   Large sample                   DF:     min       =   1,324.08
                                                        avg       =   6,513.68
                                                        max       =  29,313.78
Model F test:       Equal FMI                   F(   9,43196.6)   =       5.42
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8937375    .029854    -3.36   0.001     .8370646    .9542474
    newpolitymin1sq |   .9782259   .0063046    -3.42   0.001     .9659404    .9906677
  newincumbpowerdur |    1.04518   .0183829     2.51   0.012     1.009762    1.081841
        newgdppcthl |   .8437762   .0656912    -2.18   0.029     .7243613    .9828774
          newlnoill |   .8756681    .035901    -3.24   0.001     .8080409    .9489551
newmilexpsold10tile |    1.48164   .1372018     4.25   0.000     1.235549    1.776746
           civilwar |   .8493467   .5967244    -0.23   0.816     .2141284    3.368959
      newcivxmilexp |   .8207762   .0997107    -1.63   0.104     .6467305    1.041661
              lnpop |   1.076065   .1235058     0.64   0.523     .8591388    1.347764
              _cons |   .1804925   .2088926    -1.48   0.139     .0186543    1.746382
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. * generate urbxdeathtile10 = urbandum * deathtile10
. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 lnpop

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        266       77.10       77.10
          1 |         79       22.90      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) lnpop = success urbandum democrat antimonarch, add(30) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum lnpop success urbandum democrat antimonarch
                     , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum lnpop success urbandum democrat antimonarch ,
                     knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 lnpop success urbandum democrat antimonarch ,
                     knn(3)
             lnpop: pmm lnpop deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat antimonarch ,
                     knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
             lnpop: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
             lnpop |        299           46        46 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch lnpop if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.0252
                                                Largest FMI       =     0.0998
DF adjustment:   Large sample                   DF:     min       =   2,949.35
                                                        avg       = 116,517.40
                                                        max       = 333,924.13
Model F test:       Equal FMI                   F(   7,266383.7)  =       8.09
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.407074    .142628     3.37   0.001     1.153452    1.716463
       urbandum |   18.83056   19.54858     2.83   0.005     2.461448    144.0575
    deathtile10 |   1.332438   .1553404     2.46   0.014     1.060253    1.674496
urbxdeathtile10 |   .6011843   .0823439    -3.72   0.000     .4596399    .7863169
       democrat |   2.707506    .951261     2.83   0.005     1.359893    5.390562
    antimonarch |   2.430287   1.106171     1.95   0.051     .9959158    5.930515
          lnpop |   .8904569   .0824518    -1.25   0.210     .7426417    1.067693
          _cons |   .0033885   .0040133    -4.80   0.000     .0003325    .0345277
---------------------------------------------------------------------------------

. *       RESULT:  Not significant.
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta, clear

. * Bivariate relationships
. * Population size, t-1
. logit success lnpop if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        243
                                                LR chi2(1)        =       0.03
                                                Prob > chi2       =     0.8517
Log likelihood = -166.16962                     Pseudo R2         =     0.0001

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       lnpop |   .9839457   .0852222    -0.19   0.852     .8303215    1.165993
       _cons |   .8846404   .7224891    -0.15   0.881     .1784791    4.384764
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Population size, by urban/rural
. logit success i.urbandum##c.lnpop if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        243
                                                LR chi2(3)        =       6.39
                                                Prob > chi2       =     0.0941
Log likelihood = -162.99148                     Pseudo R2         =     0.0192

----------------------------------------------------------------------------------
         success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
        urbandum |
            yes  |   2.208146   3.761555     0.47   0.642     .0783455    62.23594
           lnpop |    .973817    .138778    -0.19   0.852     .7365008    1.287602
                 |
urbandum#c.lnpop |
            yes  |   .9898406   .1799629    -0.06   0.955     .6931199    1.413586
                 |
           _cons |   .6155503   .8147515    -0.37   0.714     .0459833    8.239988
----------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp lnpop if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        193
                                                LR chi2(9)        =      60.48
                                                Prob > chi2       =     0.0000
Log likelihood =  -101.9147                     Pseudo R2         =     0.2288

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9020725   .0336443    -2.76   0.006     .8384834    .9704841
    newpolitymin1sq |   .9757692   .0073093    -3.27   0.001     .9615479    .9902008
  newincumbpowerdur |   1.061428   .0228155     2.77   0.006      1.01764    1.107101
        newgdppcthl |   .8101248   .0723988    -2.36   0.018     .6799582    .9652095
          newlnoill |   .8490697   .0412615    -3.37   0.001     .7719306    .9339174
newmilexpsold10tile |    1.56923   .1604769     4.41   0.000     1.284218    1.917496
           civilwar |    1.83485   1.486089     0.75   0.454     .3751389    8.974476
      newcivxmilexp |   .7368106   .1050054    -2.14   0.032     .5572477    .9742344
              lnpop |   1.103648   .1490334     0.73   0.465     .8470064    1.438051
              _cons |   .1370481   .1865557    -1.46   0.144       .00951    1.974988
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant and changes no signs or patterns of significance
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10  democrat antimonarch lnpop if startyear>1899, or
>  nolog

Logistic regression                             Number of obs     =        264
                                                LR chi2(7)        =      41.08
                                                Prob > chi2       =     0.0000
Log likelihood = -157.68704                     Pseudo R2         =     0.1152

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.324396   .1331305     2.79   0.005     1.087561    1.612806
       urbandum |   6.706125   7.043428     1.81   0.070     .8559662    52.53959
    deathtile10 |   1.231546   .1449968     1.77   0.077     .9777637    1.551198
urbxdeathtile10 |   .6736827   .0942542    -2.82   0.005     .5121118    .8862291
       democrat |   2.451615   .9165727     2.40   0.016     1.178197    5.101367
    antimonarch |   2.039206   .9944724     1.46   0.144     .7840621    5.303614
          lnpop |   .9327413   .0879766    -0.74   0.460     .7753102     1.12214
          _cons |   .0110535   .0131901    -3.78   0.000      .001066    .1146121
---------------------------------------------------------------------------------

. *       RESULT:  Not significant.
. 
. 
. ***********************
. * Economic growth, t-1
. ***********************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newincumbage newgdppcthl newlnoill newmilexp
> sold10tile newcivxmilexp gdpgrow1yr 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        145       50.35       50.35
          1 |        143       49.65      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) gdpgrow1yr = success civilwar, add(50) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp gdpgrow1yr success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp gdpgrow1yr success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp gdpgrow1yr success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp gdpgrow1yr success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newmilexpsold10tile newcivxmilexp gdpgrow1yr success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq newcivxmilexp gdpgrow1yr success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile gdpgrow1yr success civilwar , knn(3)
        gdpgrow1yr: pmm gdpgrow1yr newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=50 .........10.........20.........30.........40.........50 done

Multivariate imputation                     Imputations =       50
Chained equations                                 added =       50
Imputed: m=1 through m=50                       updated =        0

Initialization: monotone                     Iterations =      500
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
        gdpgrow1yr: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
        gdpgrow1yr |        165          123       123 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success gdpgrow1yr if startyear>1899

Imputations (50):
  .........10.........20.........30.........40.........50 done

Multiple-imputation estimates                   Imputations       =         50
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.3384
                                                Largest FMI       =     0.4070
DF adjustment:   Large sample                   DF:     min       =     301.61
                                                        avg       =   2,772.89
                                                        max       =   5,244.18
Model F test:       Equal FMI                   F(   1,  301.6)   =       0.36
Within VCE type:          OIM                   Prob > F          =     0.5479

------------------------------------------------------------------------------
     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  gdpgrow1yr |    .996378   .0060103    -0.60   0.548     .9846205    1.008276
       _cons |   .6348158   .0885195    -3.26   0.001     .4829784    .8343874
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.gdpgrow1yr if startyear>1899

Imputations (50):
  .........10.........20.........30.........40.........50 done

Multiple-imputation estimates                   Imputations       =         50
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.2070
                                                Largest FMI       =     0.3438
DF adjustment:   Large sample                   DF:     min       =     422.17
                                                        avg       =   4,739.93
                                                        max       =  13,103.62
Model F test:       Equal FMI                   F(   3, 3005.5)   =       4.24
Within VCE type:          OIM                   Prob > F          =     0.0053

---------------------------------------------------------------------------------------
              success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
             urbandum |
                 yes  |   2.655361   .7857839     3.30   0.001     1.486653    4.742829
           gdpgrow1yr |   .9922358   .0117977    -0.66   0.512     .9693152    1.015698
                      |
urbandum#c.gdpgrow1yr |
                 yes  |   1.006211   .0129061     0.48   0.629     .9811928    1.031868
                      |
                _cons |   .3520927   .0863753    -4.26   0.000     .2176638     .569545
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp gdpgrow1yr if startyear>1899

Imputations (50):
  .........10.........20.........30.........40.........50 done

Multiple-imputation estimates                   Imputations       =         50
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1240
                                                Largest FMI       =     0.3311
DF adjustment:   Large sample                   DF:     min       =     455.02
                                                        avg       =  12,857.20
                                                        max       =  52,273.13
Model F test:       Equal FMI                   F(   9,30757.3)   =       5.28
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8914677   .0290104    -3.53   0.000      .836375    .9501895
    newpolitymin1sq |   .9783308   .0064828    -3.31   0.001     .9657005    .9911264
  newincumbpowerdur |   1.047057   .0183229     2.63   0.009     1.011753    1.083593
        newgdppcthl |   .8493341   .0681429    -2.04   0.042     .7257352    .9939828
          newlnoill |   .8827177   .0322414    -3.42   0.001     .8217329    .9482284
newmilexpsold10tile |   1.481173   .1385198     4.20   0.000     1.232935    1.779392
           civilwar |   .8892956   .6308747    -0.17   0.869     .2212062    3.575157
      newcivxmilexp |   .8144922   .0992567    -1.68   0.092     .6413171     1.03443
         gdpgrow1yr |   .9957973   .0065301    -0.64   0.521     .9830467    1.008713
              _cons |   .3515713   .1654574    -2.22   0.026     .1397363    .8845405
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 gdpgrow1yr 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        158       45.80       45.80
          1 |        187       54.20      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) gdpgrow1yr = success urbandum democrat antimonarch, add(50) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum gdpgrow1yr success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum gdpgrow1yr success urbandum democrat antimonarch
                     , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 gdpgrow1yr success urbandum democrat antimonarch
                     , knn(3)
        gdpgrow1yr: pmm gdpgrow1yr deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat antimonarch
                     , knn(3)

Performing chained iterations:
  imputing m=1 through m=50 .........10.........20.........30.........40.........50 done

Multivariate imputation                     Imputations =       50
Chained equations                                 added =       50
Imputed: m=1 through m=50                       updated =        0

Initialization: monotone                     Iterations =      500
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
        gdpgrow1yr: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
        gdpgrow1yr |        179          166       166 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch gdpgrow1yr if startyear>1899

Imputations (50):
  .........10.........20.........30.........40.........50 done

Multiple-imputation estimates                   Imputations       =         50
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.0654
                                                Largest FMI       =     0.2849
DF adjustment:   Large sample                   DF:     min       =     613.81
                                                        avg       = 251,451.45
                                                        max       = 547,309.73
Model F test:       Equal FMI                   F(   7,71412.7)   =       7.87
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.365745   .1245145     3.42   0.001     1.142239    1.632987
       urbandum |   18.65629   19.34675     2.82   0.005     2.444099    142.4072
    deathtile10 |   1.335615   .1561422     2.48   0.013     1.062111    1.679549
urbxdeathtile10 |   .5957489   .0815377    -3.78   0.000     .4555777    .7790476
       democrat |   2.800404   .9854616     2.93   0.003     1.405018     5.58161
    antimonarch |   2.547335   1.168733     2.04   0.042     1.036443    6.260758
     gdpgrow1yr |   .9925979    .005948    -1.24   0.216     .9809856    1.004348
          _cons |   .0017464   .0019504    -5.69   0.000     .0001956    .0155881
---------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Bivariate relationships
. * Economic growth, t-1
. logit success gdpgrow1yr if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        165
                                                LR chi2(1)        =       0.89
                                                Prob > chi2       =     0.3461
Log likelihood = -112.31725                     Pseudo R2         =     0.0039

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  gdpgrow1yr |   .9945096   .0059753    -0.92   0.360     .9828669     1.00629
       _cons |   .8050755   .1376988    -1.27   0.205     .5757694    1.125705
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Economic growth, by urban/rural
. logit success i.urbandum##c.gdpgrow1yr if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        165
                                                LR chi2(3)        =       2.73
                                                Prob > chi2       =     0.4352
Log likelihood = -111.39624                     Pseudo R2         =     0.0121

---------------------------------------------------------------------------------------
              success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
             urbandum |
                 yes  |   1.185252   .4448351     0.45   0.651     .5679994    2.473283
           gdpgrow1yr |   .9850697   .0125211    -1.18   0.237      .960832    1.009919
                      |
urbandum#c.gdpgrow1yr |
                 yes  |   1.013269   .0145463     0.92   0.359     .9851561    1.042184
                      |
                _cons |   .7293514    .229382    -1.00   0.316     .3937611    1.350955
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp gdpgrow1yr if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        146
                                                LR chi2(9)        =      41.81
                                                Prob > chi2       =     0.0000
Log likelihood = -78.922344                     Pseudo R2         =     0.2094

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9117632   .0374405    -2.25   0.024     .8412564    .9881792
    newpolitymin1sq |   .9830457   .0081164    -2.07   0.038     .9672658    .9990829
  newincumbpowerdur |   1.058093   .0275931     2.17   0.030      1.00537     1.11358
        newgdppcthl |   .7787402   .0852514    -2.28   0.022     .6283601    .9651094
          newlnoill |   .8621433   .0456524    -2.80   0.005     .7771528    .9564285
newmilexpsold10tile |   1.478132   .1769641     3.26   0.001     1.168978    1.869048
           civilwar |   1.405351   1.326958     0.36   0.719     .2208333    8.943447
      newcivxmilexp |   .7531146   .1272389    -1.68   0.093     .5408171     1.04875
         gdpgrow1yr |   .9946064   .0068282    -0.79   0.431     .9813131     1.00808
              _cons |   .4848786   .2912597    -1.21   0.228     .1493923    1.573757
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10  democrat antimonarch gdpgrow1yr if startyear>189
> 9, or nolog

Logistic regression                             Number of obs     =        158
                                                LR chi2(7)        =      26.07
                                                Prob > chi2       =     0.0005
Log likelihood = -94.947913                     Pseudo R2         =     0.1207

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.245184   .1493229     1.83   0.067     .9843676    1.575107
       urbandum |   6.691898   9.928534     1.28   0.200     .3653014    122.5878
    deathtile10 |   1.275894   .2048674     1.52   0.129     .9314059    1.747793
urbxdeathtile10 |   .6149186   .1198573    -2.49   0.013     .4196672    .9010113
       democrat |   2.851396   1.374976     2.17   0.030     1.108149    7.336969
    antimonarch |   3.029341   1.981871     1.69   0.090     .8403708    10.92007
     gdpgrow1yr |    .992545   .0066037    -1.12   0.261      .979686    1.005573
          _cons |      .0121   .0181788    -2.94   0.003     .0006367    .2299405
---------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. 
. ************************
. * Youth bulge (under 15)
. ************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newincumbage newgdppcthl newlnoill newmilexp
> sold10tile newcivxmilexp percunder15 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        130       45.14       45.14
          1 |        158       54.86      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) percunder15 = success civilwar, add(60) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp percunder15 success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp percunder15 success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp percunder15 success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp percunder15 success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newmilexpsold10tile newcivxmilexp percunder15 success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq newcivxmilexp percunder15 success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile percunder15 success civilwar , knn(3)
       percunder15: pmm percunder15 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
       percunder15: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
       percunder15 |        147          141       141 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success percunder15 if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1467
                                                Largest FMI       =     0.2269
DF adjustment:   Large sample                   DF:     min       =   1,159.41
                                                        avg       =   1,200.20
                                                        max       =   1,240.99
Model F test:       Equal FMI                   F(   1, 1159.4)   =       1.72
Within VCE type:          OIM                   Prob > F          =     0.1905

------------------------------------------------------------------------------
     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 percunder15 |     .98058   .0146814    -1.31   0.191     .9521939    1.009812
       _cons |   1.330202    .807669     0.47   0.638     .4041869    4.377768
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.percunder15 if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1912
                                                Largest FMI       =     0.3340
DF adjustment:   Large sample                   DF:     min       =     536.87
                                                        avg       =     606.47
                                                        max       =     671.11
Model F test:       Equal FMI                   F(   3, 4157.4)   =       4.38
Within VCE type:          OIM                   Prob > F          =     0.0044

----------------------------------------------------------------------------------------
               success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
              urbandum |
                  yes  |   16.26723   36.24857     1.25   0.211     .2046938    1292.774
           percunder15 |   1.036634   .0496274     0.75   0.453     .9435896    1.138853
                       |
urbandum#c.percunder15 |
                  yes  |   .9611063   .0483123    -0.79   0.430      .870776    1.060807
                       |
                 _cons |   .0647805   .1396091    -1.27   0.205     .0009397    4.465797
----------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp percunder15 if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1397
                                                Largest FMI       =     0.3898
DF adjustment:   Large sample                   DF:     min       =     394.42
                                                        avg       =   7,466.98
                                                        max       =  30,793.68
Model F test:       Equal FMI                   F(   9,30145.3)   =       5.13
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8946711   .0289225    -3.44   0.001     .8397348    .9532014
    newpolitymin1sq |   .9775039   .0063396    -3.51   0.000     .9651531    .9900127
  newincumbpowerdur |   1.046765   .0185091     2.58   0.010     1.011108    1.083679
        newgdppcthl |   .8539893      .0816    -1.65   0.099     .7080681    1.029982
          newlnoill |   .8849534   .0325274    -3.33   0.001     .8234402    .9510619
newmilexpsold10tile |   1.485136   .1439606     4.08   0.000     1.227934    1.796212
           civilwar |    .967179   .6871031    -0.05   0.963     .2401267     3.89559
      newcivxmilexp |    .797728   .1024236    -1.76   0.079     .6200949    1.026246
        percunder15 |   1.006061   .0281551     0.22   0.829     .9522034    1.062965
              _cons |   .2653375   .3730425    -0.94   0.346     .0167423    4.205163
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 percunder15 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        149       43.19       43.19
          1 |        196       56.81      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) percunder15 = success urbandum democrat antimonarch, add(60) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum percunder15 success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum percunder15 success urbandum democrat
                     antimonarch , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 percunder15 success urbandum democrat
                     antimonarch , knn(3)
       percunder15: pmm percunder15 deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat
                     antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
       percunder15: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
       percunder15 |        157          188       188 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch percunder15 if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.1375
                                                Largest FMI       =     0.4557
DF adjustment:   Large sample                   DF:     min       =     288.91
                                                        avg       =  22,642.99
                                                        max       =  57,541.65
Model F test:       Equal FMI                   F(   7,22919.7)   =       7.60
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.457476   .1454201     3.78   0.000     1.198541    1.772352
       urbandum |   19.00881   20.05269     2.79   0.005     2.404226    150.2915
    deathtile10 |   1.261745   .1518496     1.93   0.053     .9966114    1.597413
urbxdeathtile10 |   .6011446   .0839389    -3.64   0.000     .4572144    .7903837
       democrat |   2.834374   1.024755     2.88   0.004     1.395412    5.757204
    antimonarch |   2.115232   1.000577     1.58   0.113     .8369302    5.345971
    percunder15 |   1.043756   .0240379     1.86   0.064     .9975011    1.092157
          _cons |   .0002019   .0003328    -5.16   0.000     7.95e-06    .0051276
---------------------------------------------------------------------------------

. *       RESULT:  Statistically significant at the .05 level, urbandum grows insignificant, no changes in signs o
> r other patterns of significance
. 
. * Multiple imputation, impact on combined reduced model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newgdppcthl newlnoill newincumbpowerdur newmilexpsold10tile ne
> wcivxmilexp lnparticnum deathtile10 urbxdeathtile10 percunder15

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        123       42.71       42.71
          1 |        165       57.29      100.00
------------+-----------------------------------
      Total |        288      100.00

. * Impute
. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (pmm, knn(3)) percunde
> r15 = success civilwar urbandum democrat antimonarch, add(60) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
       deathtile10: truncreg deathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 lnparticnum newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
       lnparticnum: pmm lnparticnum newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 newmilexpsold10tile newcivxmilexp percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newcivxmilexp percunder15 success
                     civilwar urbandum democrat antimonarch , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile percunder15 success civilwar
                     urbandum democrat antimonarch , knn(3)
       percunder15: pmm percunder15 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
       percunder15: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
       lnparticnum |        267           21        21 |       288
       deathtile10 |        275           13        13 |       288
    urbxdeathti~10 |        275           13        13 |       288
       percunder15 |        147          141       141 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile percunder15 if st
> artyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0973
                                                Largest FMI       =     0.3039
DF adjustment:   Large sample                   DF:     min       =     648.14
                                                        avg       =  22,916.00
                                                        max       =  81,813.75
Model F test:       Equal FMI                   F(  11,72787.1)   =       5.54
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.718837    .213313     4.36   0.000      1.34769    2.192195
           urbandum |   33.71287   44.36116     2.67   0.008     2.556906     444.505
        deathtile10 |   1.242136   .1878783     1.43   0.152     .9234592    1.670786
    urbxdeathtile10 |   .5755816   .0982592    -3.24   0.001     .4119018    .8043037
      newpolitymin1 |   .8918086   .0316167    -3.23   0.001     .8319332    .9559933
    newpolitymin1sq |   .9785601   .0066422    -3.19   0.001     .9656253    .9916682
  newincumbpowerdur |   1.038099    .020445     1.90   0.058     .9987886    1.078956
        newgdppcthl |   .7815736   .0817715    -2.36   0.019     .6366092    .9595483
          newlnoill |   .8910288   .0363498    -2.83   0.005     .8225564    .9652011
newmilexpsold10tile |   1.297587    .099665     3.39   0.001     1.116095    1.508594
        percunder15 |   1.013159   .0290918     0.46   0.649      .957614    1.071926
              _cons |   .0001192   .0002465    -4.37   0.000     2.07e-06    .0068691
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Youth bulge (under 15)--cases severely reduced
. * Bivariate relationships
. logit success percunder15 if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        147
                                                LR chi2(1)        =       0.05
                                                Prob > chi2       =     0.8206
Log likelihood = -101.86352                     Pseudo R2         =     0.0003

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 percunder15 |   .9962457   .0165262    -0.23   0.821     .9643759    1.029169
       _cons |   1.139749   .7498603     0.20   0.842     .3138983    4.138369
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Youth bulge (under 15), by urban/rural
. logit success i.urbandum##c.percunder15 if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        147
                                                LR chi2(3)        =       5.71
                                                Prob > chi2       =     0.1266
Log likelihood = -99.034265                     Pseudo R2         =     0.0280

----------------------------------------------------------------------------------------
               success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
              urbandum |
                  yes  |     32.895   104.6769     1.10   0.272     .0643394    16818.34
           percunder15 |    1.07354   .0724513     1.05   0.293     .9405293    1.225362
                       |
urbandum#c.percunder15 |
                  yes  |   .9437981   .0665239    -0.82   0.412     .8220191    1.083618
                       |
                 _cons |     .02421   .0749643    -1.20   0.229      .000056    10.46336
----------------------------------------------------------------------------------------

. *       RESULT:  Marginally significant for rural revolutions
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp percunder15 if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        130
                                                LR chi2(9)        =      42.07
                                                Prob > chi2       =     0.0000
Log likelihood = -69.076628                     Pseudo R2         =     0.2334

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9151862   .0401354    -2.02   0.043     .8398082    .9973299
    newpolitymin1sq |   .9773472   .0088419    -2.53   0.011     .9601701    .9948315
  newincumbpowerdur |   1.063275   .0273983     2.38   0.017     1.010909    1.118354
        newgdppcthl |   .8050622   .1008645    -1.73   0.084     .6297733     1.02914
          newlnoill |   .8447869   .0455484    -3.13   0.002     .7600687    .9389478
newmilexpsold10tile |   1.348355   .1998306     2.02   0.044     1.008448    1.802832
           civilwar |   .3681047   .4594621    -0.80   0.423     .0318798    4.250372
      newcivxmilexp |   .8736544    .172244    -0.69   0.493     .5936388    1.285751
        percunder15 |    1.00751   .0325898     0.23   0.817     .9456181    1.073453
              _cons |   .9713008   1.643427    -0.02   0.986     .0352479    26.76544
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch percunder15 if startyear>189
> 9, or nolog

Logistic regression                             Number of obs     =        149
                                                LR chi2(7)        =      23.14
                                                Prob > chi2       =     0.0016
Log likelihood = -91.623476                     Pseudo R2         =     0.1121

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.420326   .1974457     2.52   0.012      1.08158    1.865165
       urbandum |   4.307609   6.685658     0.94   0.347      .205647    90.22983
    deathtile10 |   1.077513   .1838383     0.44   0.662     .7712514     1.50539
urbxdeathtile10 |   .7304054   .1414202    -1.62   0.105      .499753    1.067512
       democrat |   1.688193   .8494231     1.04   0.298     .6297026    4.525938
    antimonarch |   .7799284   .5647111    -0.34   0.731     .1886852    3.223826
    percunder15 |   1.047761   .0249212     1.96   0.050     1.000038    1.097763
          _cons |   .0018234   .0036889    -3.12   0.002     .0000346    .0961475
---------------------------------------------------------------------------------

. *       RESULT:  Significant, and all variables grow insignificant except lnparticnum 
. * Impact on combined reduced model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpowerdur n
> ewgdppcthl newlnoill newmilexpsold10tile percunder15 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        123
                                                LR chi2(11)       =      53.63
                                                Prob > chi2       =     0.0000
Log likelihood = -58.342302                     Pseudo R2         =     0.3149

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.347609   .2432931     1.65   0.098     .9459989    1.919717
           urbandum |   55.10644   115.3566     1.92   0.055     .9106399     3334.71
        deathtile10 |   1.099139   .2478671     0.42   0.675     .7064774    1.710044
    urbxdeathtile10 |   .5376871   .1463961    -2.28   0.023     .3153355    .9168249
      newpolitymin1 |   .9533896    .046386    -0.98   0.327      .866675     1.04878
    newpolitymin1sq |   .9901073   .0093834    -1.05   0.294     .9718859     1.00867
  newincumbpowerdur |   1.068377   .0318662     2.22   0.027     1.007711    1.132696
        newgdppcthl |   .6579737   .1052272    -2.62   0.009     .4809269    .9001977
          newlnoill |   .8804025   .0524721    -2.14   0.033     .7833387    .9894936
newmilexpsold10tile |   1.215164   .1448662     1.63   0.102     .9619631     1.53501
        percunder15 |   1.003625   .0414999     0.09   0.930     .9254951     1.08835
              _cons |   .0093047    .026538    -1.64   0.101     .0000348    2.491246
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. 
. *****************************
. * Youth bulge (aged 20 to 39)
. *****************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp youthpercl 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        143       49.65       49.65
          1 |        145       50.35      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) youthpercl = success civilwar, add(60) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp youthpercl success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp youthpercl success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp youthpercl success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp youthpercl success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newmilexpsold10tile newcivxmilexp youthpercl success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq newcivxmilexp youthpercl success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile youthpercl success civilwar , knn(3)
        youthpercl: pmm youthpercl newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
        youthpercl: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
        youthpercl |        165          123       123 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success youthpercl if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1524
                                                Largest FMI       =     0.2348
DF adjustment:   Large sample                   DF:     min       =   1,083.27
                                                        avg       =   1,096.44
                                                        max       =   1,109.61
Model F test:       Equal FMI                   F(   1, 1083.3)   =       0.16
Within VCE type:          OIM                   Prob > F          =     0.6895

------------------------------------------------------------------------------
     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  youthpercl |   1.017366   .0438292     0.40   0.689     .9349008    1.107105
       _cons |   .3707302   .4632854    -0.79   0.427     .0319291    4.304562
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.youthpercl if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1597
                                                Largest FMI       =     0.2933
DF adjustment:   Large sample                   DF:     min       =     695.65
                                                        avg       =     787.28
                                                        max       =     891.91
Model F test:       Equal FMI                   F(   3, 5602.7)   =       4.67
Within VCE type:          OIM                   Prob > F          =     0.0029

---------------------------------------------------------------------------------------
              success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
             urbandum |
                 yes  |   .0594072   .2043794    -0.82   0.412     .0000694    50.84153
           youthpercl |   .8731126   .0996801    -1.19   0.235      .697786    1.092492
                      |
urbandum#c.youthpercl |
                 yes  |   1.149767   .1418189     1.13   0.258     .9025468    1.464705
                      |
                _cons |   13.80552   43.44458     0.83   0.404     .0286275    6657.655
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp youthpercl if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1186
                                                Largest FMI       =     0.2532
DF adjustment:   Large sample                   DF:     min       =     931.97
                                                        avg       =   7,083.64
                                                        max       =  32,909.91
Model F test:       Equal FMI                   F(   9,40994.4)   =       5.27
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8879396   .0297385    -3.55   0.000     .8315132    .9481951
    newpolitymin1sq |   .9754443   .0064663    -3.75   0.000     .9628483     .988205
  newincumbpowerdur |    1.04923   .0186829     2.70   0.007     1.013242    1.086495
        newgdppcthl |   .8562641   .0687325    -1.93   0.053     .7315977    1.002174
          newlnoill |   .9005094   .0341582    -2.76   0.006     .8359829    .9700163
newmilexpsold10tile |   1.508087   .1531391     4.05   0.000     1.235601    1.840663
           civilwar |   .9007982   .6411344    -0.15   0.883     .2230698    3.637594
      newcivxmilexp |   .7970479    .099745    -1.81   0.070     .6235609    1.018803
         youthpercl |   .9110329   .0543668    -1.56   0.119     .8103568    1.024217
              _cons |   4.565618   7.455142     0.93   0.353     .1854268    112.4156
-------------------------------------------------------------------------------------

. 
. * Impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 youthpercl 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        167       48.41       48.41
          1 |        178       51.59      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) youthpercl = success urbandum democrat antimonarch, add(50) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum youthpercl success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum youthpercl success urbandum democrat antimonarch
                     , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 youthpercl success urbandum democrat antimonarch
                     , knn(3)
        youthpercl: pmm youthpercl deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat antimonarch
                     , knn(3)

Performing chained iterations:
  imputing m=1 through m=50 .........10.........20.........30.........40.........50 done

Multivariate imputation                     Imputations =       50
Chained equations                                 added =       50
Imputed: m=1 through m=50                       updated =        0

Initialization: monotone                     Iterations =      500
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
        youthpercl: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
        youthpercl |        175          170       170 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch youthpercl if startyear>1899

Imputations (50):
  .........10.........20.........30.........40.........50 done

Multiple-imputation estimates                   Imputations       =         50
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.1113
                                                Largest FMI       =     0.3935
DF adjustment:   Large sample                   DF:     min       =     322.55
                                                        avg       =  24,168.59
                                                        max       =  66,774.43
Model F test:       Equal FMI                   F(   7,27255.9)   =       7.68
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.404568   .1366879     3.49   0.000     1.160573    1.699858
       urbandum |   16.16753   17.09766     2.63   0.009      2.03438    128.4858
    deathtile10 |   1.286258   .1536307     2.11   0.035     1.017786    1.625549
urbxdeathtile10 |   .6118585   .0855264    -3.51   0.000     .4652259    .8047077
       democrat |   3.027552   1.092053     3.07   0.002     1.492978    6.139453
    antimonarch |   2.749172   1.283623     2.17   0.030     1.100912    6.865172
     youthpercl |   .9123716   .0500608    -1.67   0.096     .8190141    1.016371
          _cons |   .0208751   .0378285    -2.14   0.033     .0005949    .7324518
---------------------------------------------------------------------------------

. *       RESULT:  Marginally significant at the .10 level
. 
. * Impact on combined reduced model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newgdppcthl newlnoill newincumbpowerdur newmilexpsold10tile ne
> wcivxmilexp lnparticnum deathtile10 urbxdeathtile10 youthpercl 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        136       47.22       47.22
          1 |        152       52.78      100.00
------------+-----------------------------------
      Total |        288      100.00

. * Impute
. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (pmm, knn(3)) youthper
> cl = success civilwar urbandum democrat antimonarch, add(60) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
       deathtile10: truncreg deathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 lnparticnum newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
       lnparticnum: pmm lnparticnum newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 newmilexpsold10tile newcivxmilexp youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newcivxmilexp youthpercl success
                     civilwar urbandum democrat antimonarch , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile youthpercl success civilwar
                     urbandum democrat antimonarch , knn(3)
        youthpercl: pmm youthpercl newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
        youthpercl: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
       lnparticnum |        267           21        21 |       288
       deathtile10 |        275           13        13 |       288
    urbxdeathti~10 |        275           13        13 |       288
        youthpercl |        165          123       123 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile youthpercl if sta
> rtyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0917
                                                Largest FMI       =     0.2692
DF adjustment:   Large sample                   DF:     min       =     825.17
                                                        avg       =  13,771.77
                                                        max       =  35,391.92
Model F test:       Equal FMI                   F(  11,82820.5)   =       5.57
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.722322   .2155831     4.34   0.000     1.347589    2.201259
           urbandum |   39.03479   52.32376     2.73   0.006      2.82106    540.1215
        deathtile10 |   1.241936   .1888725     1.42   0.154     .9218168    1.673222
    urbxdeathtile10 |   .5676823   .0985442    -3.26   0.001     .4039608    .7977584
      newpolitymin1 |   .8808441   .0323448    -3.46   0.001     .8196643    .9465905
    newpolitymin1sq |   .9772203   .0067642    -3.33   0.001     .9640493    .9905713
  newincumbpowerdur |   1.040072   .0208723     1.96   0.050     .9999543    1.081799
        newgdppcthl |   .7616973   .0675925    -3.07   0.002     .6400928    .9064042
          newlnoill |   .9093967   .0395714    -2.18   0.029     .8350416    .9903727
newmilexpsold10tile |   1.338457   .1029678     3.79   0.000     1.151049    1.556377
         youthpercl |   .9073572   .0595007    -1.48   0.139     .7977707    1.031997
              _cons |   .0025723    .005773    -2.66   0.008     .0000315     .209961
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Youth bulge (aged 20 to 39)--highly reduced sample
. * Bivariate
. logit success youthpercl if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        165
                                                LR chi2(1)        =       0.18
                                                Prob > chi2       =     0.6691
Log likelihood =  -114.1294                     Pseudo R2         =     0.0008

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  youthpercl |   .9808593   .0444374    -0.43   0.670     .8975184    1.071939
       _cons |   1.615587   2.152481     0.36   0.719      .118647    21.99906
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Youth bulge (aged 20 to 39), by urban/rural
. logit success i.urbandum##c.youthpercl if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        165
                                                LR chi2(3)        =       4.87
                                                Prob > chi2       =     0.1816
Log likelihood = -111.78615                     Pseudo R2         =     0.0213

---------------------------------------------------------------------------------------
              success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
             urbandum |
                 yes  |   .0215846   .0861659    -0.96   0.337     8.63e-06    53.96889
           youthpercl |   .8317978   .1107569    -1.38   0.167     .6407332    1.079837
                      |
urbandum#c.youthpercl |
                 yes  |   1.174147   .1681018     1.12   0.262     .8868628    1.554492
                      |
                _cons |   103.2189   377.9027     1.27   0.205     .0789516    134945.3
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp youthpercl if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        143
                                                LR chi2(9)        =      43.41
                                                Prob > chi2       =     0.0000
Log likelihood = -77.385574                     Pseudo R2         =     0.2190

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8947395   .0375773    -2.65   0.008     .8240391    .9715058
    newpolitymin1sq |   .9736658   .0084997    -3.06   0.002     .9571483    .9904683
  newincumbpowerdur |   1.051372   .0254592     2.07   0.039     1.002638    1.102474
        newgdppcthl |   .8031135    .078467    -2.24   0.025     .6631497     .972618
          newlnoill |   .8786349   .0438103    -2.59   0.009     .7968307    .9688373
newmilexpsold10tile |   1.363636   .1922653     2.20   0.028     1.034388    1.797684
           civilwar |   .4588943    .536546    -0.67   0.505     .0463954    4.538894
      newcivxmilexp |   .8675234   .1580862    -0.78   0.435     .6069729    1.239918
         youthpercl |   .9231137   .0655607    -1.13   0.260     .8031595    1.060983
              _cons |   11.81146    24.1234     1.21   0.227     .2156915    646.8061
-------------------------------------------------------------------------------------

. *       RESULT: Not significant
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10  democrat antimonarch youthpercl if startyear>189
> 9, or nolog

Logistic regression                             Number of obs     =        167
                                                LR chi2(7)        =      21.04
                                                Prob > chi2       =     0.0037
Log likelihood = -105.23192                     Pseudo R2         =     0.0909

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.362796   .1697147     2.49   0.013     1.067645     1.73954
       urbandum |   2.157826   3.135637     0.53   0.597     .1250537    37.23373
    deathtile10 |   1.118492   .1780103     0.70   0.482     .8187714     1.52793
urbxdeathtile10 |   .7819054   .1429824    -1.35   0.179     .5463866    1.118944
       democrat |   2.225714   1.021274     1.74   0.081     .9055103     5.47073
    antimonarch |   1.318128   .9205335     0.40   0.692     .3353582    5.180913
     youthpercl |   .9152017   .0492369    -1.65   0.100     .8236127    1.016976
          _cons |   .1952656   .3943653    -0.81   0.419     .0037283    10.22687
---------------------------------------------------------------------------------

. *       RESULT:  Marginally significant, turns all variables grow insignificant except lnparticnum 
. * Impact on combined reduced model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpowerdur n
> ewgdppcthl newlnoill newmilexpsold10tile youthpercl if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        136
                                                LR chi2(11)       =      53.94
                                                Prob > chi2       =     0.0000
Log likelihood = -67.282704                     Pseudo R2         =     0.2862

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.473683   .2455508     2.33   0.020     1.063099    2.042841
           urbandum |   23.58895   46.60024     1.60   0.110     .4910782    1133.096
        deathtile10 |   1.108479   .2404535     0.47   0.635     .7245749     1.69579
    urbxdeathtile10 |   .6001191   .1523389    -2.01   0.044       .36489    .9869906
      newpolitymin1 |   .9132401   .0419503    -1.98   0.048     .8346116    .9992762
    newpolitymin1sq |    .980852   .0090183    -2.10   0.035     .9633348    .9986878
  newincumbpowerdur |   1.051721   .0288889     1.84   0.066     .9965968    1.109894
        newgdppcthl |    .695013   .0803683    -3.15   0.002     .5540687    .8718108
          newlnoill |   .9208089   .0525345    -1.45   0.148     .8233914    1.029752
newmilexpsold10tile |   1.225223   .1400684     1.78   0.076     .9792762     1.53294
         youthpercl |   .9196826   .0695847    -1.11   0.268     .7929296    1.066698
              _cons |   .0539562   .1488317    -1.06   0.290     .0002422    12.02203
-------------------------------------------------------------------------------------

. *       RESULT: Not significant
. 
. 
. ********************
. * Percent mountains
. ********************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp lnmtnest 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        201       69.79       69.79
          1 |         87       30.21      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnmtnest = success civilwar, add(40) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq lnmtnest
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq lnmtnest
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq lnmtnest
                     newmilexpsold10tile newcivxmilexp success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq lnmtnest
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 lnmtnest
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
          lnmtnest: pmm lnmtnest newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq lnmtnest newcivxmilexp success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     lnmtnest newmilexpsold10tile success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=40 .........10.........20.........30.........40 done

Multivariate imputation                     Imputations =       40
Chained equations                                 added =       40
Imputed: m=1 through m=40                       updated =        0

Initialization: monotone                     Iterations =      400
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
          lnmtnest: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
          lnmtnest |        250           38        38 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success lnmtnest if startyear>1899

Imputations (40):
  .........10.........20.........30.........40 done

Multiple-imputation estimates                   Imputations       =         40
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0467
                                                Largest FMI       =     0.0858
DF adjustment:   Large sample                   DF:     min       =   5,344.66
                                                        avg       =   7,013.29
                                                        max       =   8,681.92
Model F test:       Equal FMI                   F(   1, 5344.7)   =       0.05
Within VCE type:          OIM                   Prob > F          =     0.8196

------------------------------------------------------------------------------
     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    lnmtnest |   1.022562   .1000537     0.23   0.820     .8440813    1.238784
       _cons |   .5771685   .1534172    -2.07   0.039     .3427782    .9718339
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.lnmtnest if startyear>1899

Imputations (40):
  .........10.........20.........30.........40 done

Multiple-imputation estimates                   Imputations       =         40
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0545
                                                Largest FMI       =     0.1497
DF adjustment:   Large sample                   DF:     min       =   1,763.38
                                                        avg       =   2,712.02
                                                        max       =   3,866.26
Model F test:       Equal FMI                   F(   3,23900.2)   =       5.04
Within VCE type:          OIM                   Prob > F          =     0.0017

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
           urbandum |
               yes  |   2.743506   1.680663     1.65   0.100     .8254657    9.118274
           lnmtnest |   1.036551   .1968635     0.19   0.850      .714197      1.5044
                    |
urbandum#c.lnmtnest |
               yes  |   1.018924   .2276828     0.08   0.933     .6574304    1.579189
                    |
              _cons |   .2947024   .1545481    -2.33   0.020      .105387    .8241007
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp lnmtnest if startyear>1899

Imputations (40):
  .........10.........20.........30.........40 done

Multiple-imputation estimates                   Imputations       =         40
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0723
                                                Largest FMI       =     0.1459
DF adjustment:   Large sample                   DF:     min       =   1,856.22
                                                        avg       =   8,555.43
                                                        max       =  20,648.70
Model F test:       Equal FMI                   F(   9,65388.2)   =       5.52
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8931724    .029367    -3.44   0.001     .8374116    .9526462
    newpolitymin1sq |   .9775208   .0062489    -3.56   0.000     .9653465    .9898486
  newincumbpowerdur |   1.046578   .0184609     2.58   0.010     1.011011    1.083397
        newgdppcthl |    .846308   .0658173    -2.15   0.032     .7266524    .9856669
          newlnoill |    .882216   .0326806    -3.38   0.001      .820429    .9486563
newmilexpsold10tile |   1.487713   .1365643     4.33   0.000     1.242603    1.781173
           civilwar |   .8675204   .5888753    -0.21   0.834     .2292544    3.282779
      newcivxmilexp |   .8172866    .094854    -1.74   0.082      .650954    1.026121
           lnmtnest |   1.092757   .1300745     0.75   0.456     .8653439    1.379934
              _cons |   .2767588   .1624568    -2.19   0.029     .0875688    .8746891
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 lnmtnest 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        269       77.97       77.97
          1 |         76       22.03      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) lnmtnest = success urbandum democrat antimonarch, add(30) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum lnmtnest success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum lnmtnest success urbandum democrat antimonarch ,
                     knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 lnmtnest success urbandum democrat antimonarch ,
                     knn(3)
          lnmtnest: pmm lnmtnest deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat antimonarch ,
                     knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
          lnmtnest: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
          lnmtnest |        303           42        42 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch lnmtnest if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.0232
                                                Largest FMI       =     0.0769
DF adjustment:   Large sample                   DF:     min       =   4,957.13
                                                        avg       = 170,366.80
                                                        max       = 638,012.45
Model F test:       Equal FMI                   F(   7,309435.1)  =       8.07
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.343498   .1225182     3.24   0.001     1.123555    1.606497
       urbandum |   20.13886    20.8894     2.89   0.004     2.636912    153.8063
    deathtile10 |   1.335137   .1568319     2.46   0.014     1.060566    1.680792
urbxdeathtile10 |   .5905579   .0808817    -3.85   0.000     .4515254     .772401
       democrat |   2.806139    .982131     2.95   0.003     1.413166    5.572181
    antimonarch |   2.453849    1.10728     1.99   0.047     1.013327    5.942188
       lnmtnest |   .9582558   .1000954    -0.41   0.683     .7808319    1.175995
          _cons |   .0020831   .0023362    -5.51   0.000     .0002313     .018765
---------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Percent mountains
. * Bivariate
. logit success lnmtnest if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        250
                                                LR chi2(1)        =       0.06
                                                Prob > chi2       =     0.8101
Log likelihood = -169.36567                     Pseudo R2         =     0.0002

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    lnmtnest |    1.02408   .1014416     0.24   0.810     .8433674    1.243514
       _cons |   .6615148   .1799816    -1.52   0.129     .3881048    1.127535
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Percent mountains, by urban/rural
. logit success i.urbandum##c.lnmtnest if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        250
                                                LR chi2(3)        =       9.75
                                                Prob > chi2       =     0.0208
Log likelihood = -164.51827                     Pseudo R2         =     0.0288

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
           urbandum |
               yes  |   2.172116   1.360713     1.24   0.216     .6362844    7.415061
           lnmtnest |   1.028523   .1968638     0.15   0.883     .7067899     1.49671
                    |
urbandum#c.lnmtnest |
               yes  |   1.028454   .2314567     0.12   0.901     .6616359    1.598639
                    |
              _cons |   .3832935   .2063185    -1.78   0.075     .1334591    1.100816
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp lnmtnest if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        201
                                                LR chi2(9)        =      60.39
                                                Prob > chi2       =     0.0000
Log likelihood = -106.06275                     Pseudo R2         =     0.2216

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9162844   .0334318    -2.40   0.017     .8530474    .9842092
    newpolitymin1sq |   .9775444   .0073231    -3.03   0.002     .9632963    .9920033
  newincumbpowerdur |   1.066392   .0225377     3.04   0.002     1.023121    1.111493
        newgdppcthl |   .8204143   .0725114    -2.24   0.025     .6899232    .9755863
          newlnoill |   .8663697   .0384958    -3.23   0.001     .7941114    .9452029
newmilexpsold10tile |   1.530531     .15502     4.20   0.000     1.254955    1.866621
           civilwar |   1.367181   1.062608     0.40   0.687     .2980236    6.271935
      newcivxmilexp |   .7465149   .1028871    -2.12   0.034      .569801    .9780336
           lnmtnest |   1.004042   .1392095     0.03   0.977     .7651268    1.317559
              _cons |   .3328982   .2151971    -1.70   0.089     .0937702    1.181838
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant and changes no signs or patterns of significance
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch lnmtnest if startyear>1899, 
> or nolog

Logistic regression                             Number of obs     =        267
                                                LR chi2(7)        =      45.56
                                                Prob > chi2       =     0.0000
Log likelihood = -155.71798                     Pseudo R2         =     0.1276

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.256799   .1169725     2.46   0.014     1.047232    1.508303
       urbandum |   9.956721   10.74171     2.13   0.033     1.201727    82.49482
    deathtile10 |   1.295264   .1553916     2.16   0.031      1.02386    1.638612
urbxdeathtile10 |   .6466929    .092026    -3.06   0.002     .4892941    .8547247
       democrat |   2.949864   1.132283     2.82   0.005     1.390192     6.25935
    antimonarch |   2.484264    1.24045     1.82   0.068     .9336267    6.610315
       lnmtnest |   .9288386   .1029395    -0.67   0.505     .7474893    1.154185
          _cons |   .0070298     .00811    -4.30   0.000     .0007327    .0674428
---------------------------------------------------------------------------------

. *       RESULT:  Not significant 
. 
. *******************************
. * Avg total years of schooling
. *******************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp avgtotalyrsschool 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        110       38.19       38.19
          1 |        178       61.81      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) avgtotalyrsschool = success civilwar, add(70) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp avgtotalyrsschool success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp avgtotalyrsschool success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp avgtotalyrsschool success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp avgtotalyrsschool success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newmilexpsold10tile newcivxmilexp avgtotalyrsschool success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq newcivxmilexp avgtotalyrsschool success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile avgtotalyrsschool success civilwar , knn(3)
    avgtotalyrss~l: pmm avgtotalyrsschool newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=70 .........10.........20.........30.........40.........50.........60.........70 done

Multivariate imputation                     Imputations =       70
Chained equations                                 added =       70
Imputed: m=1 through m=70                       updated =        0

Initialization: monotone                     Iterations =      700
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
    avgtotalyrss~l: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
    avgtotalyrss~l |        121          167       167 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success avgtotalyrsschool if startyear>1899

Imputations (70):
  .........10.........20.........30.........40.........50.........60.........70 done

Multiple-imputation estimates                   Imputations       =         70
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.2804
                                                Largest FMI       =     0.3614
DF adjustment:   Large sample                   DF:     min       =     535.45
                                                        avg       =     717.59
                                                        max       =     899.74
Model F test:       Equal FMI                   F(   1,  535.4)   =       0.01
Within VCE type:          OIM                   Prob > F          =     0.9174

-----------------------------------------------------------------------------------
          success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
avgtotalyrsschool |   1.005868   .0566974     0.10   0.917     .9004365    1.123645
            _cons |   .5953102   .1501158    -2.06   0.040     .3629207    .9765059
-----------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.avgtotalyrsschool if startyear>
> 1899

Imputations (70):
  .........10.........20.........30.........40.........50.........60.........70 done

Multiple-imputation estimates                   Imputations       =         70
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.3507
                                                Largest FMI       =     0.4968
DF adjustment:   Large sample                   DF:     min       =     283.60
                                                        avg       =     424.55
                                                        max       =     675.19
Model F test:       Equal FMI                   F(   3, 2000.9)   =       3.93
Within VCE type:          OIM                   Prob > F          =     0.0083

----------------------------------------------------------------------------------------------
                     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-----------------------------+----------------------------------------------------------------
                    urbandum |
                        yes  |   1.875454   1.115367     1.06   0.291     .5834077    6.028936
           avgtotalyrsschool |   .8127208   .1505101    -1.12   0.264     .5644549    1.170182
                             |
urbandum#c.avgtotalyrsschool |
                        yes  |   1.194424   .2282065     0.93   0.353     .8202771    1.739228
                             |
                       _cons |    .562128   .2906718    -1.11   0.266     .2033858    1.553638
----------------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp avgtotalyrsschool if startyear>1899

Imputations (70):
  .........10.........20.........30.........40.........50.........60.........70 done

Multiple-imputation estimates                   Imputations       =         70
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1925
                                                Largest FMI       =     0.5100
DF adjustment:   Large sample                   DF:     min       =     269.12
                                                        avg       =   4,032.28
                                                        max       =   9,550.18
Model F test:       Equal FMI                   F(   9,20440.8)   =       5.05
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9062282   .0310697    -2.87   0.004     .8473133    .9692396
    newpolitymin1sq |   .9763904   .0064754    -3.60   0.000     .9637765    .9891693
  newincumbpowerdur |   1.050584   .0192806     2.69   0.007     1.013462    1.089067
        newgdppcthl |   .8861299   .0743647    -1.44   0.150      .751711    1.044585
          newlnoill |   .8776179   .0336557    -3.40   0.001     .8140631    .9461345
newmilexpsold10tile |   1.578851   .1751723     4.12   0.000     1.269882    1.962994
           civilwar |     1.0243   .7278286     0.03   0.973     .2543104     4.12563
      newcivxmilexp |   .7690962   .0989003    -2.04   0.041     .5976584    .9897108
  avgtotalyrsschool |   .8533559   .0900342    -1.50   0.134     .6932942    1.050371
              _cons |   .4540323   .2254555    -1.59   0.112     .1714893    1.202089
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 avgtotalyrsschool 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        122       35.36       35.36
          1 |        223       64.64      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) avgtotalyrsschool = success urbandum democrat antimonarch, add(70) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum avgtotalyrsschool success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum avgtotalyrsschool success urbandum democrat
                     antimonarch , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 avgtotalyrsschool success urbandum democrat
                     antimonarch , knn(3)
    avgtotalyrss~l: pmm avgtotalyrsschool deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat
                     antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=70 .........10.........20.........30.........40.........50.........60.........70 done

Multivariate imputation                     Imputations =       70
Chained equations                                 added =       70
Imputed: m=1 through m=70                       updated =        0

Initialization: monotone                     Iterations =      700
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    avgtotalyrss~l: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
    avgtotalyrss~l |        129          216       216 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch avgtotalyrsschool if startyear>1899

Imputations (70):
  .........10.........20.........30.........40.........50.........60.........70 done

Multiple-imputation estimates                   Imputations       =         70
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.1713
                                                Largest FMI       =     0.4596
DF adjustment:   Large sample                   DF:     min       =     331.26
                                                        avg       =   8,541.17
                                                        max       =  15,196.98
Model F test:       Equal FMI                   F(   7,18911.7)   =       7.81
Within VCE type:          OIM                   Prob > F          =     0.0000

-----------------------------------------------------------------------------------
          success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
      lnparticnum |   1.481502   .1529955     3.81   0.000     1.209938    1.814019
         urbandum |   21.23478   22.98034     2.82   0.005      2.54547    177.1444
      deathtile10 |   1.240172   .1524627     1.75   0.080     .9745925    1.578123
  urbxdeathtile10 |   .5952572   .0851248    -3.63   0.000     .4497475    .7878446
         democrat |   2.755135   1.048008     2.66   0.008     1.307125    5.807228
      antimonarch |   1.655587   .8172971     1.02   0.307     .6290689    4.357183
avgtotalyrsschool |   .8013997   .0653212    -2.72   0.007     .6826755    .9407712
            _cons |   .0023286   .0026973    -5.23   0.000     .0002404    .0225522
-----------------------------------------------------------------------------------

. *       RESULT:  Statistically significant at the .01 level, but negative (seems opposite of what one would expe
> ct theoretically--it should not reduce the chance of opposition victory)
. 
. * Multiple imputation, impact on combined reduced model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newgdppcthl newlnoill newincumbpowerdur newmilexpsold10tile ne
> wcivxmilexp lnparticnum deathtile10 urbxdeathtile10 avgtotalyrsschool 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        104       36.11       36.11
          1 |        184       63.89      100.00
------------+-----------------------------------
      Total |        288      100.00

. * Impute
. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (pmm, knn(3)) avgtotal
> yrsschool = success civilwar urbandum democrat antimonarch, add(70) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
       deathtile10: truncreg deathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 lnparticnum newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
       lnparticnum: pmm lnparticnum newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 newmilexpsold10tile newcivxmilexp avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newcivxmilexp avgtotalyrsschool
                     success civilwar urbandum democrat antimonarch , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile avgtotalyrsschool success
                     civilwar urbandum democrat antimonarch , knn(3)
    avgtotalyrss~l: pmm avgtotalyrsschool newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=70 .........10.........20.........30.........40.........50.........60.........70 done

Multivariate imputation                     Imputations =       70
Chained equations                                 added =       70
Imputed: m=1 through m=70                       updated =        0

Initialization: monotone                     Iterations =      700
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    avgtotalyrss~l: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
       lnparticnum |        267           21        21 |       288
       deathtile10 |        275           13        13 |       288
    urbxdeathti~10 |        275           13        13 |       288
    avgtotalyrss~l |        121          167       167 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile avgtotalyrsschool
>  if startyear>1899

Imputations (70):
  .........10.........20.........30.........40.........50.........60.........70 done

Multiple-imputation estimates                   Imputations       =         70
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1328
                                                Largest FMI       =     0.4080
DF adjustment:   Large sample                   DF:     min       =     420.29
                                                        avg       =   9,136.09
                                                        max       =  26,984.89
Model F test:       Equal FMI                   F(  11,49042.8)   =       5.44
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.749403   .2233326     4.38   0.000     1.362093    2.246844
           urbandum |   36.83398   50.47558     2.63   0.009     2.509915    540.5529
        deathtile10 |   1.242154   .1940937     1.39   0.165     .9144403    1.687314
    urbxdeathtile10 |   .5780554    .102678    -3.09   0.002     .4080905    .8188087
      newpolitymin1 |   .9026984   .0332124    -2.78   0.005     .8398766    .9702192
    newpolitymin1sq |   .9782923   .0068061    -3.15   0.002     .9650396    .9917269
  newincumbpowerdur |   1.038404    .020857     1.88   0.061     .9983163    1.080102
        newgdppcthl |   .7979646   .0755747    -2.38   0.017     .6627473    .9607698
          newlnoill |   .8843902   .0368255    -2.95   0.003     .8150774    .9595973
newmilexpsold10tile |   1.350898   .1066996     3.81   0.000     1.157075    1.577189
  avgtotalyrsschool |   .8577469   .0868034    -1.52   0.130     .7030228    1.046523
              _cons |   .0002231   .0003504    -5.35   0.000     .0000103    .0048507
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Avg total years of schooling--highly reduced sample
. * Bivariate
. logit success avgtotalyrsschool if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        121
                                                LR chi2(1)        =       0.60
                                                Prob > chi2       =     0.4371
Log likelihood = -83.465467                     Pseudo R2         =     0.0036

-----------------------------------------------------------------------------------
          success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
avgtotalyrsschool |   .9512613   .0613672    -0.77   0.439     .8382772    1.079474
            _cons |   1.145397   .3838891     0.41   0.685      .593839    2.209241
-----------------------------------------------------------------------------------

. *       RESULT: Not significant
. * Avg total years of schooling, by urban/rural
. logit success i.urbandum##c.avgtotalyrsschool if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        121
                                                LR chi2(3)        =       7.25
                                                Prob > chi2       =     0.0644
Log likelihood = -80.143548                     Pseudo R2         =     0.0433

----------------------------------------------------------------------------------------------
                     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------------+----------------------------------------------------------------
                    urbandum |
                        yes  |   1.393617   1.228368     0.38   0.707     .2476702    7.841753
           avgtotalyrsschool |   .6960652   .1924239    -1.31   0.190     .4048921    1.196632
                             |
urbandum#c.avgtotalyrsschool |
                        yes  |    1.30261   .3730607     0.92   0.356     .7430785    2.283464
                             |
                       _cons |   1.371855   1.042726     0.42   0.677     .3092597    6.085457
----------------------------------------------------------------------------------------------

. *       RESULT: Not significant
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp avgtotalyrsschool if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        110
                                                LR chi2(9)        =      35.70
                                                Prob > chi2       =     0.0000
Log likelihood = -58.231505                     Pseudo R2         =     0.2346

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9376016   .0489403    -1.23   0.217     .8464238    1.038601
    newpolitymin1sq |   .9723971   .0101445    -2.68   0.007     .9527162    .9924846
  newincumbpowerdur |   1.053007   .0291195     1.87   0.062     .9974531    1.111655
        newgdppcthl |   .8815782   .1005342    -1.11   0.269     .7050027    1.102379
          newlnoill |   .8454089   .0496004    -2.86   0.004     .7535751    .9484339
newmilexpsold10tile |   1.335623   .2268767     1.70   0.088     .9574011    1.863262
           civilwar |   .1907805   .2692454    -1.17   0.240     .0120018    3.032637
      newcivxmilexp |   .9264044   .2001929    -0.35   0.724     .6065391    1.414954
  avgtotalyrsschool |   .8469707   .1030128    -1.37   0.172     .6673306    1.074969
              _cons |   3.073102   3.173926     1.09   0.277     .4059276    23.26511
-------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 democrat antimonarch avgtotalyrsschool if startyear>1899, or nolo
> g

Logistic regression                             Number of obs     =        122
                                                LR chi2(6)        =      19.73
                                                Prob > chi2       =     0.0031
Log likelihood =  -74.68357                     Pseudo R2         =     0.1167

-----------------------------------------------------------------------------------
          success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
      lnparticnum |   1.387927   .1950222     2.33   0.020     1.053808    1.827981
         urbandum |    .457679   .3197323    -1.12   0.263       .11639    1.799725
      deathtile10 |   .8576264   .0804965    -1.64   0.102     .7135176    1.030841
         democrat |   2.715466   1.461827     1.86   0.064     .9453956    7.799648
      antimonarch |   .9750069    .762435    -0.03   0.974     .2105648    4.514707
avgtotalyrsschool |   .8156387   .0719557    -2.31   0.021     .6861274    .9695962
            _cons |    .150545   .2226829    -1.28   0.201     .0082906    2.733666
-----------------------------------------------------------------------------------

. *       RESULT: Is statistically significant, but switches signs from negative to positive
. * Impact on combined reduced model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 newpolitymin1 newpolitymin1sq newincumbpowerdur n
> ewgdppcthl newlnoill newmilexpsold10tile avgtotalyrsschool if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        104
                                                LR chi2(11)       =      39.22
                                                Prob > chi2       =     0.0000
Log likelihood = -52.456308                     Pseudo R2         =     0.2721

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |     1.3671   .2470752     1.73   0.084     .9593196    1.948218
           urbandum |   14.76062   37.03705     1.07   0.283     .1079661    2018.004
        deathtile10 |   1.005758   .2805008     0.02   0.984     .5822346    1.737358
    urbxdeathtile10 |   .6766853   .2111857    -1.25   0.211     .3670596     1.24749
      newpolitymin1 |   .9440737   .0526975    -1.03   0.303     .8462378    1.053221
    newpolitymin1sq |   .9814615   .0107441    -1.71   0.087     .9606277    1.002747
  newincumbpowerdur |   1.050184   .0334532     1.54   0.124     .9866214    1.117841
        newgdppcthl |   .7730689   .1029133    -1.93   0.053     .5955301    1.003535
          newlnoill |   .9040943   .0554096    -1.65   0.100     .8017626    1.019487
newmilexpsold10tile |   1.158185   .1526209     1.11   0.265     .8945612    1.499498
  avgtotalyrsschool |   .9116239   .1197876    -0.70   0.481     .7046406    1.179407
              _cons |   .0286853   .0844521    -1.21   0.228     .0000895    9.197642
-------------------------------------------------------------------------------------

. *       RESULT: Not significant
. 
. *****************************************
. * Military personnel per 1000 population
. *****************************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp milper1000

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        212       73.61       73.61
          1 |         76       26.39      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) milper1000 = success civilwar, add(30) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq milper1000
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq milper1000
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq milper1000
                     newmilexpsold10tile newcivxmilexp success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq milper1000
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 milper1000
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
        milper1000: pmm milper1000 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq milper1000 newcivxmilexp success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     milper1000 newmilexpsold10tile success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
        milper1000: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
        milper1000 |        245           43        43 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success milper1000 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1097
                                                Largest FMI       =     0.1815
DF adjustment:   Large sample                   DF:     min       =     898.60
                                                        avg       =   4,747.50
                                                        max       =   8,596.40
Model F test:       Equal FMI                   F(   1,  898.6)   =       0.08
Within VCE type:          OIM                   Prob > F          =     0.7793

------------------------------------------------------------------------------
     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  milper1000 |   1.002849   .0101799     0.28   0.779     .9830676    1.023028
       _cons |   .5952162   .0870705    -3.55   0.000     .4468278    .7928834
------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.milper1000 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0915
                                                Largest FMI       =     0.1238
DF adjustment:   Large sample                   DF:     min       =   1,921.82
                                                        avg       =   5,538.83
                                                        max       =  10,881.25
Model F test:       Equal FMI                   F(   3, 6760.3)   =       4.82
Within VCE type:          OIM                   Prob > F          =     0.0024

---------------------------------------------------------------------------------------
              success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
             urbandum |
                 yes  |   2.735457   .9267036     2.97   0.003     1.408038    5.314293
           milper1000 |   .9840729   .0386285    -0.41   0.683     .9111659    1.062814
                      |
urbandum#c.milper1000 |
                 yes  |   1.012622   .0414783     0.31   0.759     .9344562    1.097326
                      |
                _cons |    .347999   .0966256    -3.80   0.000     .2019326    .5997214
---------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp milper1000 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0915
                                                Largest FMI       =     0.2218
DF adjustment:   Large sample                   DF:     min       =     603.67
                                                        avg       =   6,743.14
                                                        max       =  19,865.09
Model F test:       Equal FMI                   F(   9,30988.5)   =       5.46
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8929023   .0291435    -3.47   0.001     .8375536    .9519086
    newpolitymin1sq |   .9770172   .0063261    -3.59   0.000     .9646905    .9895014
  newincumbpowerdur |   1.046578    .018597     2.56   0.010     1.010754    1.083672
        newgdppcthl |   .8378614   .0661575    -2.24   0.025     .7177214    .9781117
          newlnoill |   .8839529   .0324918    -3.36   0.001     .8225046    .9499919
newmilexpsold10tile |   1.492078   .1354322     4.41   0.000       1.2488    1.782749
           civilwar |   .9081135   .6361362    -0.14   0.891     .2298337    3.588116
      newcivxmilexp |   .8121476   .0968135    -1.75   0.081     .6428033    1.026105
         milper1000 |   1.002596   .0127504     0.20   0.839     .9778655    1.027952
              _cons |   .3404887   .1603526    -2.29   0.022     .1352581    .8571214
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 milper1000

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        217       62.90       62.90
          1 |        128       37.10      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) milper1000 = success urbandum democrat antimonarch, add(30) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum milper1000 success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum milper1000 success urbandum democrat antimonarch
                     , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 milper1000 success urbandum democrat antimonarch
                     , knn(3)
        milper1000: pmm milper1000 deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat antimonarch
                     , knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
        milper1000: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
        milper1000 |        246           99        99 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch milper1000 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.0308
                                                Largest FMI       =     0.1165
DF adjustment:   Large sample                   DF:     min       =   2,168.77
                                                        avg       = 147,117.01
                                                        max       = 447,333.57
Model F test:       Equal FMI                   F(   7,176625.6)  =       7.99
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.341712   .1207439     3.27   0.001     1.124714    1.600575
       urbandum |   19.55307   20.28278     2.87   0.004     2.559992    149.3452
    deathtile10 |   1.331174   .1554876     2.45   0.014     1.058789    1.673633
urbxdeathtile10 |   .5935415   .0815485    -3.80   0.000       .45342    .7769652
       democrat |    2.78853   .9757242     2.93   0.003     1.404539    5.536266
    antimonarch |   2.446659   1.113244     1.97   0.049     1.002929    5.968656
     milper1000 |   1.000797   .0121218     0.07   0.948     .9773056    1.024853
          _cons |   .0019368   .0021475    -5.63   0.000     .0002204    .0170183
---------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Military personnel per 1000 population
. * Bivariate
. logit success milper1000 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        245
                                                LR chi2(1)        =       0.25
                                                Prob > chi2       =     0.6201
Log likelihood = -162.52384                     Pseudo R2         =     0.0008

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  milper1000 |    1.00495   .0099717     0.50   0.619     .9855945    1.024685
       _cons |   .5882301   .0905467    -3.45   0.001     .4350319    .7953777
------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * Military personnel per 1000 population, by urban/rural
. logit success i.urbandum##c.milper1000 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        245
                                                LR chi2(3)        =      13.51
                                                Prob > chi2       =     0.0037
Log likelihood = -155.89096                     Pseudo R2         =     0.0415

---------------------------------------------------------------------------------------
              success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------+----------------------------------------------------------------
             urbandum |
                 yes  |   2.598915   .9091892     2.73   0.006      1.30922     5.15907
           milper1000 |   .9867477   .0386693    -0.34   0.734     .9137948    1.065525
                      |
urbandum#c.milper1000 |
                 yes  |   1.011629   .0410965     0.28   0.776     .9342044     1.09547
                      |
                _cons |   .3592153    .102423    -3.59   0.000     .2054243     .628142
---------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp milper1000 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        212
                                                LR chi2(9)        =      65.61
                                                Prob > chi2       =     0.0000
Log likelihood = -106.65963                     Pseudo R2         =     0.2352

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9040665   .0333836    -2.73   0.006     .8409474    .9719231
    newpolitymin1sq |   .9755699   .0072809    -3.31   0.001     .9614034    .9899452
  newincumbpowerdur |   1.064713   .0228886     2.92   0.004     1.020784    1.110532
        newgdppcthl |   .8761749   .0773065    -1.50   0.134     .7370344    1.041583
          newlnoill |    .868069   .0364367    -3.37   0.001      .799513    .9425034
newmilexpsold10tile |   1.510233   .1513035     4.12   0.000     1.240983    1.837901
           civilwar |   1.272654   .9886492     0.31   0.756     .2776278    5.833888
      newcivxmilexp |    .772479    .104376    -1.91   0.056     .5927531    1.006699
         milper1000 |   .9996568   .0129843    -0.03   0.979     .9745292    1.025432
              _cons |   .2852233   .1481433    -2.42   0.016     .1030567    .7893943
-------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. 
. 
. ****************************
. * V-Dem civil society index
. ****************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp newvdcivsocmin1 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        227       78.82       78.82
          1 |         61       21.18      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) newvdcivsocmin1 = success civilwar, add(30) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newvdcivsocmin1 newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newvdcivsocmin1 newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newvdcivsocmin1 newpolitymin1
                     newpolitymin1sq newmilexpsold10tile newcivxmilexp success civilwar , ll(0)
    newvdcivsocm~1: pmm newvdcivsocmin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newvdcivsocmin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newvdcivsocmin1 newpolitymin1
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newvdcivsocmin1
                     newpolitymin1 newpolitymin1sq newcivxmilexp success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newvdcivsocmin1 newpolitymin1
                     newpolitymin1sq newmilexpsold10tile success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
    newvdcivsocm~1: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
    newvdcivsocm~1 |        279            9         9 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success newvdcivsocmin1 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0064
                                                Largest FMI       =     0.0122
DF adjustment:   Large sample                   DF:     min       = 195,476.15
                                                        avg       = 273,447.21
                                                        max       = 351,418.27
Model F test:       Equal FMI                   F(   1,195476.2)  =       6.05
Within VCE type:          OIM                   Prob > F          =     0.0139

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
newvdcivsocmin1 |    .304547   .1471507    -2.46   0.014     .1181324    .7851262
          _cons |   1.002086   .2329583     0.01   0.993     .6353644    1.580474
---------------------------------------------------------------------------------

. *       RESULT:  Significant at the .05 level, but negative (i.e., thicker civil society lowers probability of o
> pposition victory?)  Does not make theoretical sense.
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.newvdcivsocmin1 if startyear>18
> 99

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0043
                                                Largest FMI       =     0.0066
DF adjustment:   Large sample                   DF:     min       = 672,165.49
                                                        avg       = 1352294.66
                                                        max       = 1938940.56
Model F test:       Equal FMI                   F(   3, 2.5e+06)  =       6.69
Within VCE type:          OIM                   Prob > F          =     0.0002

--------------------------------------------------------------------------------------------
                   success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
---------------------------+----------------------------------------------------------------
                  urbandum |
                      yes  |   2.776522   1.386322     2.05   0.041     1.043505    7.387679
           newvdcivsocmin1 |   .3350012   .2755372    -1.33   0.184     .0668245     1.67941
                           |
urbandum#c.newvdcivsocmin1 |
                      yes  |   .9939712   1.023052    -0.01   0.995     .1322109    7.472747
                           |
                     _cons |    .518693   .2084275    -1.63   0.102     .2359772    1.140121
--------------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp newvdcivsocmin1 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0745
                                                Largest FMI       =     0.2004
DF adjustment:   Large sample                   DF:     min       =     738.16
                                                        avg       =  10,293.41
                                                        max       =  57,653.01
Model F test:       Equal FMI                   F(   9,46901.2)   =       5.55
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8928757   .0403404    -2.51   0.012      .817184    .9755785
    newpolitymin1sq |   .9773219   .0064058    -3.50   0.000     .9648362    .9899692
  newincumbpowerdur |   1.045888   .0183284     2.56   0.010     1.010574    1.082436
        newgdppcthl |   .8446537   .0663478    -2.15   0.032     .7241103     .985264
          newlnoill |   .8847584   .0323597    -3.35   0.001     .8235507    .9505151
newmilexpsold10tile |   1.484885   .1396195     4.20   0.000     1.234623    1.785876
           civilwar |   .8765801   .6233779    -0.19   0.853     .2171661    3.538272
      newcivxmilexp |   .8162462   .1008001    -1.64   0.101     .6405186    1.040185
    newvdcivsocmin1 |   1.085702   .9824876     0.09   0.928     .1842093     6.39897
              _cons |   .3372951   .2179787    -1.68   0.093     .0950098    1.197434
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. * generate urbxdeathtile10 = urbandum * deathtile10
. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 newvdcivsocmin1 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        249       86.46       86.46
          1 |         39       13.54      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) newvdcivsocmin1 = success urbandum democrat antimonarch, add(20) rseed(1234) force dots

Conditional models:
    newvdcivsocm~1: pmm newvdcivsocmin1 deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat
                     antimonarch , knn(3)
       deathtile10: truncreg deathtile10 newvdcivsocmin1 urbxdeathtile10 lnparticnum success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 newvdcivsocmin1 deathtile10 lnparticnum success urbandum democrat
                     antimonarch , knn(3)
       lnparticnum: pmm lnparticnum newvdcivsocmin1 deathtile10 urbxdeathtile10 success urbandum democrat
                     antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=20 .........10.........20 done

Multivariate imputation                     Imputations =       20
Chained equations                                 added =       20
Imputed: m=1 through m=20                       updated =        0

Initialization: monotone                     Iterations =      200
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    newvdcivsocm~1: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        267           21        21 |       288
       deathtile10 |        275           13        13 |       288
    urbxdeathti~10 |        275           13        13 |       288
    newvdcivsocm~1 |        279            9         9 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch newvdcivsocmin1 if startyear>1899

Imputations (20):
  .........10.........20 done

Multiple-imputation estimates                   Imputations       =         20
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0219
                                                Largest FMI       =     0.0835
DF adjustment:   Large sample                   DF:     min       =   2,771.12
                                                        avg       =  63,049.33
                                                        max       = 132,490.30
Model F test:       Equal FMI                   F(   7,221015.5)  =       7.63
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.415313   .1467105     3.35   0.001     1.154991    1.734307
       urbandum |   23.22148   28.78975     2.54   0.011     2.044353     263.769
    deathtile10 |   1.304477   .1851383     1.87   0.061     .9877049    1.722842
urbxdeathtile10 |   .5761796   .0915356    -3.47   0.001     .4220163     .786659
       democrat |   2.130649   .8048412     2.00   0.045     1.016184    4.467367
    antimonarch |   2.182871    1.07792     1.58   0.114       .82925    5.746065
newvdcivsocmin1 |   .3456995   .2024039    -1.81   0.070      .109724     1.08917
          _cons |   .0020275   .0027309    -4.60   0.000     .0001447    .0284125
---------------------------------------------------------------------------------

. *       RESULT:  Marginally significant at the .10 level, but negative (thicker civil society should aid revolut
> ionaries, not hinder them)
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * V-Dem civil society index
. * Bivariate
. logit success newvdcivsocmin1 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        279
                                                LR chi2(1)        =       6.06
                                                Prob > chi2       =     0.0139
Log likelihood = -181.73861                     Pseudo R2         =     0.0164

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
newvdcivsocmin1 |   .3069065   .1494514    -2.43   0.015     .1181693    .7970905
          _cons |   .9931655   .2343617    -0.03   0.977     .6254028    1.577188
---------------------------------------------------------------------------------

. *       RESULT: Marginally significant at .05 level, but negative (by theory, should be positive, not negative--
> probably proxying for something else)
. * V-Dem civil society index, by urban/rural
. logit success i.urbandum##c.newvdcivsocmin1 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        279
                                                LR chi2(3)        =      19.54
                                                Prob > chi2       =     0.0002
Log likelihood = -174.99792                     Pseudo R2         =     0.0529

--------------------------------------------------------------------------------------------
                   success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
---------------------------+----------------------------------------------------------------
                  urbandum |
                      yes  |   2.426617   1.224203     1.76   0.079     .9027739    6.522644
           newvdcivsocmin1 |   .3047999   .2523817    -1.43   0.151      .060145    1.544651
                           |
urbandum#c.newvdcivsocmin1 |
                      yes  |   1.177905   1.220236     0.16   0.874     .1546389    8.972257
                           |
                     _cons |   .5603258   .2272535    -1.43   0.153     .2530554    1.240697
--------------------------------------------------------------------------------------------

. *       RESULT:  not significant
. * Impact on opposition-specific model 
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch newvdcivsocmin1 if startyear
> >1899, or nolog

Logistic regression                             Number of obs     =        249
                                                LR chi2(7)        =      56.42
                                                Prob > chi2       =     0.0000
Log likelihood = -137.32867                     Pseudo R2         =     0.1704

---------------------------------------------------------------------------------
        success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.370009   .1430296     3.02   0.003     1.116497    1.681084
       urbandum |    13.4409   16.30004     2.14   0.032     1.247842    144.7761
    deathtile10 |   1.268161   .1739494     1.73   0.083     .9692097    1.659322
urbxdeathtile10 |   .6194624   .0967984    -3.06   0.002     .4560417    .8414444
       democrat |   2.314984   .9322976     2.08   0.037     1.051355    5.097374
    antimonarch |   2.529767    1.44981     1.62   0.105     .8227289    7.778651
newvdcivsocmin1 |   .3922424   .2347919    -1.56   0.118     .1213486    1.267868
          _cons |   .0037823   .0050691    -4.16   0.000     .0002735    .0523066
---------------------------------------------------------------------------------

. *       RESULT: Not significant
. 
.  
. ******************************************
. * Private ownership of the economy (V-Dem)
. ******************************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp newvdstateownmin1 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        227       78.82       78.82
          1 |         61       21.18      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) newvdstateownmin1 = success civilwar, add(30) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newvdstateownmin1 newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newvdstateownmin1 newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newvdstateownmin1 newpolitymin1
                     newpolitymin1sq newmilexpsold10tile newcivxmilexp success civilwar , ll(0)
    newvdstateow~1: pmm newvdstateownmin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newvdstateownmin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newvdstateownmin1 newpolitymin1
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newvdstateownmin1
                     newpolitymin1 newpolitymin1sq newcivxmilexp success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newvdstateownmin1 newpolitymin1
                     newpolitymin1sq newmilexpsold10tile success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
    newvdstateow~1: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
    newvdstateow~1 |        279            9         9 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success newvdstateownmin1 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0187
                                                Largest FMI       =     0.0361
DF adjustment:   Large sample                   DF:     min       =  22,329.34
                                                        avg       =  25,849.39
                                                        max       =  29,369.45
Model F test:       Equal FMI                   F(   1,22329.3)   =       0.05
Within VCE type:          OIM                   Prob > F          =     0.8315

-----------------------------------------------------------------------------------
          success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newvdstateownmin1 |   .9702251   .1378396    -0.21   0.832     .7344056    1.281767
            _cons |   .6523433   .2254741    -1.24   0.216     .3313273    1.284385
-----------------------------------------------------------------------------------

. *       RESULT:  Not significant.
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.newvdstateownmin1 if startyear>
> 1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0110
                                                Largest FMI       =     0.0155
DF adjustment:   Large sample                   DF:     min       = 121,595.84
                                                        avg       = 189,905.27
                                                        max       = 263,888.43
Model F test:       Equal FMI                   F(   3,378593.6)  =       5.27
Within VCE type:          OIM                   Prob > F          =     0.0012

----------------------------------------------------------------------------------------------
                     success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-----------------------------+----------------------------------------------------------------
                    urbandum |
                        yes  |   3.374179   2.757976     1.49   0.137     .6798459    16.74657
           newvdstateownmin1 |   1.064657      .3052     0.22   0.827     .6070139     1.86733
                             |
urbandum#c.newvdstateownmin1 |
                        yes  |   .9299928   .3084453    -0.22   0.827     .4854715    1.781539
                             |
                       _cons |   .2783112   .1978259    -1.80   0.072     .0691014    1.120919
----------------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp newvdstateownmin1 if startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0892
                                                Largest FMI       =     0.2040
DF adjustment:   Large sample                   DF:     min       =     712.17
                                                        avg       =   5,504.84
                                                        max       =  13,591.24
Model F test:       Equal FMI                   F(   9,33361.7)   =       5.45
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8754161   .0309066    -3.77   0.000     .8168744    .9381533
    newpolitymin1sq |   .9774131     .00632    -3.53   0.000     .9650986    .9898847
  newincumbpowerdur |    1.04698   .0185539     2.59   0.010     1.011236    1.083987
        newgdppcthl |   .8531272   .0668705    -2.03   0.043      .731617    .9948183
          newlnoill |   .8913093   .0329715    -3.11   0.002     .8289656    .9583416
newmilexpsold10tile |   1.494912   .1413705     4.25   0.000     1.241726    1.799722
           civilwar |   .8591014    .620547    -0.21   0.834      .208138    3.545989
      newcivxmilexp |   .8224593   .1018226    -1.58   0.115     .6449916    1.048757
  newvdstateownmin1 |   1.319311   .2476547     1.48   0.140     .9131453    1.906138
              _cons |   .1573031   .1168248    -2.49   0.013     .0366813    .6745756
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant.
. 
. * Impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 newvdstateownmin1 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        249       72.17       72.17
          1 |         96       27.83      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) newvdstateownmin1 = success urbandum democrat antimonarch, add(20) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum newvdstateownmin1 success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum newvdstateownmin1 success urbandum democrat
                     antimonarch , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 newvdstateownmin1 success urbandum democrat
                     antimonarch , knn(3)
    newvdstateow~1: pmm newvdstateownmin1 deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat
                     antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=20 .........10.........20 done

Multivariate imputation                     Imputations =       20
Chained equations                                 added =       20
Imputed: m=1 through m=20                       updated =        0

Initialization: monotone                     Iterations =      200
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    newvdstateow~1: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
    newvdstateow~1 |        279           66        66 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch vdstateownmin1 if startyear>1899

Imputations (20):
  .........10.........20 done

Multiple-imputation estimates                   Imputations       =         20
Logistic regression                             Number of obs     =        252
                                                Average RVI       =     0.0159
                                                Largest FMI       =     0.0623
DF adjustment:   Large sample                   DF:     min       =   4,954.67
                                                        avg       = 331,719.21
                                                        max       = 1196747.85
Model F test:       Equal FMI                   F(   7,408302.0)  =       5.38
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |   1.265107   .1253051     2.37   0.018     1.041832    1.536232
       urbandum |   10.97362   12.83714     2.05   0.041     1.108088     108.674
    deathtile10 |   1.315333   .1777998     2.03   0.043     1.009184    1.714357
urbxdeathtile10 |   .6014236   .0934703    -3.27   0.001     .4434947    .8155911
       democrat |   3.541409   1.346762     3.33   0.001     1.680629     7.46243
    antimonarch |   3.337942   1.792647     2.24   0.025     1.165035    9.563537
 vdstateownmin1 |   1.085017   .1849412     0.48   0.632     .7768712    1.515389
          _cons |   .0045711   .0063941    -3.85   0.000     .0002946    .0709175
---------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Private ownership of the economy (V-Dem)
. * Bivariate
. logit success newvdstateownmin1 if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        279
                                                LR chi2(1)        =       0.02
                                                Prob > chi2       =     0.9004
Log likelihood =  -184.7588                     Pseudo R2         =     0.0000

-----------------------------------------------------------------------------------
          success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
newvdstateownmin1 |   .9822706   .1403809    -0.13   0.900     .7423042    1.299811
            _cons |   .6286805   .2197862    -1.33   0.184     .3168477    1.247411
-----------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * Private ownership of the economy (V-Dem), by urban/rural
. logit success i.urbandum##c.newvdstateownmin1 if startyear>1899 & colony==0, or nolog 

Logistic regression                             Number of obs     =        279
                                                LR chi2(3)        =      14.70
                                                Prob > chi2       =     0.0021
Log likelihood = -177.41583                     Pseudo R2         =     0.0398

----------------------------------------------------------------------------------------------
                     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------------+----------------------------------------------------------------
                    urbandum |
                        yes  |   3.046774   2.505892     1.35   0.176     .6077837    15.27325
           newvdstateownmin1 |   1.058647   .3041337     0.20   0.843     .6028563    1.859041
                             |
urbandum#c.newvdstateownmin1 |
                        yes  |   .9470499   .3154879    -0.16   0.870     .4929634    1.819412
                             |
                       _cons |   .2913523   .2076482    -1.73   0.084     .0720714    1.177807
----------------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp newvdstateownmin1 if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        227
                                                LR chi2(9)        =      71.90
                                                Prob > chi2       =     0.0000
Log likelihood = -113.63651                     Pseudo R2         =     0.2403

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8978175   .0341842    -2.83   0.005     .8332567    .9673806
    newpolitymin1sq |   .9730156   .0070031    -3.80   0.000      .959386    .9868388
  newincumbpowerdur |   1.069871   .0222774     3.24   0.001     1.027087    1.114437
        newgdppcthl |    .875322   .0723212    -1.61   0.107     .7444569    1.029191
          newlnoill |    .874132   .0350674    -3.35   0.001     .8080338    .9456372
newmilexpsold10tile |   1.494702   .1459675     4.12   0.000     1.234324    1.810006
           civilwar |    1.00687   .7707703     0.01   0.993     .2245791    4.514168
      newcivxmilexp |   .8116627   .1037921    -1.63   0.103     .6317241    1.042855
  newvdstateownmin1 |   1.146239   .2492386     0.63   0.530     .7484963    1.755338
              _cons |    .223305   .1849939    -1.81   0.070     .0440283    1.132568
-------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * Impact on opposition-specific model
. logit success lnparticnum urbandum deathtile10 urbxdeathtile10 democrat antimonarch newvdstateownmin1 if startye
> ar>1899, or nolog

Logistic regression                             Number of obs     =        249
                                                LR chi2(7)        =      54.19
                                                Prob > chi2       =     0.0000
Log likelihood = -138.44309                     Pseudo R2         =     0.1637

-----------------------------------------------------------------------------------
          success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
      lnparticnum |   1.368246   .1444657     2.97   0.003     1.112475    1.682821
         urbandum |   15.68337   18.95252     2.28   0.023     1.468276    167.5217
      deathtile10 |   1.326534   .1809949     2.07   0.038     1.015263    1.733237
  urbxdeathtile10 |   .6165595   .0960224    -3.11   0.002       .45437    .8366434
         democrat |   2.683634   1.056314     2.51   0.012     1.240735    5.804535
      antimonarch |   2.658205   1.518885     1.71   0.087     .8673934    8.146309
newvdstateownmin1 |   1.086932   .1841945     0.49   0.623     .7797504    1.515128
            _cons |   .0014684   .0021651    -4.42   0.000     .0000816    .0264186
-----------------------------------------------------------------------------------

. *       RESULT: Not significant 
. 
. 
. ***********************************
. * V-Dem executive corruption index
. ***********************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp newv2x_execorr 

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        232       80.56       80.56
          1 |         56       19.44      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) newv2x_execorr = success civilwar, add(20) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newv2x_execorr newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newv2x_execorr: pmm newv2x_execorr newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newv2x_execorr newpolitymin1
                     newpolitymin1sq newmilexpsold10tile newcivxmilexp success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newv2x_execorr newgdppcthl
                     newpolitymin1 newpolitymin1sq newcivxmilexp success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     newpolitymin1sq newmilexpsold10tile success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=20 .........10.........20 done

Multivariate imputation                     Imputations =       20
Chained equations                                 added =       20
Imputed: m=1 through m=20                       updated =        0

Initialization: monotone                     Iterations =      200
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
    newv2x_execorr: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
    newv2x_execorr |        285            3         3 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success newv2x_execorr if startyear>1899

Imputations (20):
  .........10.........20 done

Multiple-imputation estimates                   Imputations       =         20
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0052
                                                Largest FMI       =     0.0101
DF adjustment:   Large sample                   DF:     min       = 185,396.95
                                                        avg       = 215,716.47
                                                        max       = 246,035.99
Model F test:       Equal FMI                   F(   1,185396.9)  =       5.42
Within VCE type:          OIM                   Prob > F          =     0.0199

--------------------------------------------------------------------------------
       success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
---------------+----------------------------------------------------------------
newv2x_execorr |   3.185637   1.585739     2.33   0.020     1.200838    8.451003
         _cons |   .3016396   .0996748    -3.63   0.000     .1578385    .5764527
--------------------------------------------------------------------------------

. *       RESULT:  Significant at the .05 level.
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.newv2x_execorr if startyear>189
> 9

Imputations (20):
  .........10.........20 done

Multiple-imputation estimates                   Imputations       =         20
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0027
                                                Largest FMI       =     0.0040
DF adjustment:   Large sample                   DF:     min       = 1205449.56
                                                        avg       = 1964043.85
                                                        max       = 3099980.78
Model F test:       Equal FMI                   F(   3, 4.0e+06)  =       7.23
Within VCE type:          OIM                   Prob > F          =     0.0001

-------------------------------------------------------------------------------------------
                  success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------------+----------------------------------------------------------------
                 urbandum |
                     yes  |   4.814124   3.734878     2.03   0.043     1.052303    22.02389
           newv2x_execorr |    6.47198   6.079199     1.99   0.047     1.026819     40.7925
                          |
urbandum#c.newv2x_execorr |
                     yes  |   .4989578   .5602325    -0.62   0.536     .0552496    4.506076
                          |
                    _cons |   .0963663   .0644266    -3.50   0.000     .0259924    .3572762
-------------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp newv2x_execorr if startyear>1899

Imputations (20):
  .........10.........20 done

Multiple-imputation estimates                   Imputations       =         20
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0733
                                                Largest FMI       =     0.1876
DF adjustment:   Large sample                   DF:     min       =     556.93
                                                        avg       =   6,839.16
                                                        max       =  18,932.25
Model F test:       Equal FMI                   F(   9,31355.8)   =       5.53
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8982111   .0291899    -3.30   0.001     .8427563     .957315
    newpolitymin1sq |    .977949   .0062825    -3.47   0.001     .9657068    .9903464
  newincumbpowerdur |    1.04667   .0187885     2.54   0.011     1.010482    1.084153
        newgdppcthl |   .8389369   .0663898    -2.22   0.027     .7183803     .979725
          newlnoill |   .8873208   .0323346    -3.28   0.001     .8261486    .9530224
newmilexpsold10tile |   1.490206   .1415698     4.20   0.000     1.236533    1.795918
           civilwar |    .943646   .6636071    -0.08   0.934     .2374117    3.750732
      newcivxmilexp |   .8073776   .0962418    -1.79   0.073     .6389597    1.020187
     newv2x_execorr |   .9420589   .6028734    -0.09   0.926     .2687289    3.302492
              _cons |   .3478589   .1961022    -1.87   0.061      .115213     1.05028
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant.
. 
. * Multiple imputation, impact on opposition-specific model
. clear

. use revolutionaryeps.dta

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed lnparticnum deathtile10 urbxdeathtile10 newv2x_execorr

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        253       73.33       73.33
          1 |         92       26.67      100.00
------------+-----------------------------------
      Total |        345      100.00

. mi impute chained (pmm, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (
> pmm, knn(3)) newv2x_execorr = success urbandum democrat antimonarch, add(20) rseed(1234) force dots

Conditional models:
       deathtile10: truncreg deathtile10 urbxdeathtile10 lnparticnum newv2x_execorr success urbandum democrat
                     antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 deathtile10 lnparticnum newv2x_execorr success urbandum democrat
                     antimonarch , knn(3)
       lnparticnum: pmm lnparticnum deathtile10 urbxdeathtile10 newv2x_execorr success urbandum democrat
                     antimonarch , knn(3)
    newv2x_execorr: pmm newv2x_execorr deathtile10 urbxdeathtile10 lnparticnum success urbandum democrat
                     antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=20 .........10.........20 done

Multivariate imputation                     Imputations =       20
Chained equations                                 added =       20
Imputed: m=1 through m=20                       updated =        0

Initialization: monotone                     Iterations =      200
                                                burn-in =       10

       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    newv2x_execorr: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
       lnparticnum |        322           23        23 |       345
       deathtile10 |        327           18        18 |       345
    urbxdeathti~10 |        327           18        18 |       345
    newv2x_execorr |        285           60        60 |       345
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  democrat antimonarch newv2x_execorr if startyear>1899

Imputations (20):
  .........10.........20 done

Multiple-imputation estimates                   Imputations       =         20
Logistic regression                             Number of obs     =        343
                                                Average RVI       =     0.0372
                                                Largest FMI       =     0.1548
DF adjustment:   Large sample                   DF:     min       =     814.72
                                                        avg       =  47,401.46
                                                        max       = 198,531.69
Model F test:       Equal FMI                   F(   7,79604.8)   =       8.51
Within VCE type:          OIM                   Prob > F          =     0.0000

---------------------------------------------------------------------------------
        success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    lnparticnum |    1.37365   .1241312     3.51   0.000     1.150657     1.63986
       urbandum |   26.13122   28.11019     3.03   0.002     3.172865    215.2127
    deathtile10 |   1.347611   .1614016     2.49   0.013     1.065651    1.704174
urbxdeathtile10 |   .5795117   .0816123    -3.87   0.000     .4397274    .7637319
       democrat |   2.212641   .8083653     2.17   0.030     1.081246    4.527905
    antimonarch |   2.248715   1.030188     1.77   0.077     .9161724    5.519397
 newv2x_execorr |   4.098064   2.365796     2.44   0.015     1.319631    12.72639
          _cons |   .0005774   .0007159    -6.01   0.000     .0000508    .0065604
---------------------------------------------------------------------------------

. *       RESULT:  Significant at the .05 level (though not clear how this fits theoretically into the opposition 
> model)
. 
. * Multiple imputation, impact on combined reduced model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newgdppcthl newlnoill newincumbpowerdur newmilexpsold10tile ne
> wcivxmilexp lnparticnum deathtile10 urbxdeathtile10 newv2x_execorr

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        211       73.26       73.26
          1 |         77       26.74      100.00
------------+-----------------------------------
      Total |        288      100.00

. * Impute
. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (pmm, knn(3)) newv2x_e
> xecorr = success civilwar urbandum democrat antimonarch, add(30) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newv2x_execorr newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)
    newv2x_execorr: pmm newv2x_execorr newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newv2x_execorr newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp
                     success civilwar urbandum democrat antimonarch , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)
       deathtile10: truncreg deathtile10 newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     newpolitymin1sq urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success
                     civilwar urbandum democrat antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)
       lnparticnum: pmm lnparticnum newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 newmilexpsold10tile newcivxmilexp success
                     civilwar urbandum democrat antimonarch , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newv2x_execorr newgdppcthl
                     newpolitymin1 newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newcivxmilexp success
                     civilwar urbandum democrat antimonarch , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newv2x_execorr newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile success
                     civilwar urbandum democrat antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=30 .........10.........20.........30 done

Multivariate imputation                     Imputations =       30
Chained equations                                 added =       30
Imputed: m=1 through m=30                       updated =        0

Initialization: monotone                     Iterations =      300
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    newv2x_execorr: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
       lnparticnum |        267           21        21 |       288
       deathtile10 |        275           13        13 |       288
    urbxdeathti~10 |        275           13        13 |       288
    newv2x_execorr |        285            3         3 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile newv2x_execorr if
>  startyear>1899

Imputations (30):
  .........10.........20.........30 done

Multiple-imputation estimates                   Imputations       =         30
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.0581
                                                Largest FMI       =     0.1591
DF adjustment:   Large sample                   DF:     min       =   1,166.56
                                                        avg       =  19,135.69
                                                        max       =  46,877.59
Model F test:       Equal FMI                   F(  11,94814.7)   =       5.73
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.720957   .2123601     4.40   0.000     1.351206    2.191889
           urbandum |    37.3839   49.64778     2.73   0.006     2.768246    504.8525
        deathtile10 |   1.270327   .1924663     1.58   0.114     .9439459    1.709557
    urbxdeathtile10 |   .5697846   .0979717    -3.27   0.001     .4067665    .7981346
      newpolitymin1 |   .8960499   .0322796    -3.05   0.002     .8349338    .9616397
    newpolitymin1sq |   .9796301   .0069183    -2.91   0.004      .966153    .9932952
  newincumbpowerdur |   1.032579   .0204597     1.62   0.106     .9932457     1.07347
        newgdppcthl |   .7649538   .0666045    -3.08   0.002     .6449337    .9073092
          newlnoill |   .8913477   .0364158    -2.82   0.005      .822749     .965666
newmilexpsold10tile |   1.287187    .097829     3.32   0.001     1.108872    1.494177
     newv2x_execorr |    1.68029    1.17154     0.74   0.457     .4284173    6.590243
              _cons |   .0001373   .0002256    -5.41   0.000     5.49e-06    .0034372
-------------------------------------------------------------------------------------

. *       RESULT:  Not significant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * V-Dem executive corruption index
. * Bivariate
. logit success v2x_execorr if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        228
                                                LR chi2(1)        =       3.00
                                                Prob > chi2       =     0.0833
Log likelihood =  -154.5588                     Pseudo R2         =     0.0096

------------------------------------------------------------------------------
     success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 v2x_execorr |   2.459511   1.291728     1.71   0.087     .8786179    6.884897
       _cons |    .437306   .1562323    -2.32   0.021     .2171123    .8808184
------------------------------------------------------------------------------

. *       RESULT: Marginally significant at the .10 level
. * Executive corruption, by urban/rural
. logit success i.urbandum##c.v2x_execorr if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        228
                                                LR chi2(3)        =       7.06
                                                Prob > chi2       =     0.0700
Log likelihood = -152.52853                     Pseudo R2         =     0.0226

----------------------------------------------------------------------------------------
               success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
              urbandum |
                  yes  |   1.724287   1.427946     0.66   0.511     .3401716    8.740194
           v2x_execorr |   2.721804   2.690807     1.01   0.311     .3920571    18.89577
                       |
urbandum#c.v2x_execorr |
                  yes  |   1.057544   1.242558     0.05   0.962     .1057259    10.57829
                       |
                 _cons |   .2799036   .2001724    -1.78   0.075     .0689086    1.136955
----------------------------------------------------------------------------------------

. *       RESULT: No interaction
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp vdstateownmin1 v2x_execorr if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        172
                                                LR chi2(10)       =      52.57
                                                Prob > chi2       =     0.0000
Log likelihood = -91.523136                     Pseudo R2         =     0.2231

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .9101105   .0402023    -2.13   0.033     .8346301    .9924171
    newpolitymin1sq |   .9802212   .0077697    -2.52   0.012     .9651106    .9955684
  newincumbpowerdur |   1.070601   .0251679     2.90   0.004     1.022392    1.121083
        newgdppcthl |   .7683891   .0772312    -2.62   0.009     .6309958    .9356984
          newlnoill |   .8829772   .0425772    -2.58   0.010     .8033495    .9704976
newmilexpsold10tile |   1.588872   .1831262     4.02   0.000     1.267603    1.991566
           civilwar |   1.893762    1.64733     0.73   0.463     .3442619    10.41745
      newcivxmilexp |   .7272994   .1115275    -2.08   0.038     .5385003    .9822917
     vdstateownmin1 |   1.251922   .3403527     0.83   0.409      .734795    2.132988
        v2x_execorr |   .6651604   .5983127    -0.45   0.650     .1140957    3.877782
              _cons |   .2232746   .2014584    -1.66   0.097     .0380898     1.30879
-------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. 
. 
. *************************************
. * Sigman/Hanson state capacity index
. *************************************
. * Multiple imputation, impact on regime-specific model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile ne
> wcivxmilexp newstatecapacity

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        142       49.31       49.31
          1 |        146       50.69      100.00
------------+-----------------------------------
      Total |        288      100.00

. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) newstatecapacity = success civilwar, add(60) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp newstatecapacity success civilwar , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp newstatecapacity success civilwar , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp newstatecapacity success civilwar , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp newstatecapacity success civilwar , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newmilexpsold10tile newcivxmilexp newstatecapacity success civilwar , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq newcivxmilexp newstatecapacity success civilwar , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newstatecapacity success civilwar , knn(3)
    newstatecapa~y: pmm newstatecapacity newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     newmilexpsold10tile newcivxmilexp success civilwar , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
    newstatecapa~y: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
    newstatecapa~y |        155          133       133 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. * Bivariate
. mi estimate, post dots eform saving(miest, replace): logit success newstatecapacity if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.2209
                                                Largest FMI       =     0.3081
DF adjustment:   Large sample                   DF:     min       =     630.50
                                                        avg       =   2,056.13
                                                        max       =   3,481.76
Model F test:       Equal FMI                   F(   1,  630.5)   =       0.70
Within VCE type:          OIM                   Prob > F          =     0.4015

----------------------------------------------------------------------------------
         success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
newstatecapacity |    .859175   .1553482    -0.84   0.402     .6023918    1.225418
           _cons |    .557283    .090433    -3.60   0.000     .4054145    .7660417
----------------------------------------------------------------------------------

. *       RESULT:  Not significant.
. * By urban/rural
. mi estimate, post dots eform saving(miest, replace): logit success i.urbandum##c.newstatecapacity if startyear>1
> 899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1871
                                                Largest FMI       =     0.2718
DF adjustment:   Large sample                   DF:     min       =     809.12
                                                        avg       =   1,665.17
                                                        max       =   2,449.23
Model F test:       Equal FMI                   F(   3, 4339.9)   =       5.59
Within VCE type:          OIM                   Prob > F          =     0.0008

---------------------------------------------------------------------------------------------
                    success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------------------+----------------------------------------------------------------
                   urbandum |
                       yes  |   5.404648   2.670381     3.41   0.001     2.051113    14.24116
           newstatecapacity |   .4594809   .1893995    -1.89   0.060      .204585    1.031956
                            |
urbandum#c.newstatecapacity |
                       yes  |   1.644947   .7233631     1.13   0.258      .694383    3.896772
                            |
                      _cons |   .1539278   .0723083    -3.98   0.000     .0612546    .3868082
---------------------------------------------------------------------------------------------

. *       RESULT:  Marginally significant for rural revolutions, but not significant for urban
. * Impact on regime-specific model
. mi estimate, post dots eform saving(miest, replace): logit success newpolitymin1 newpolitymin1sq newincumbpowerd
> ur newgdppcthl newlnoill newmilexpsold10tile civilwar newcivxmilexp newstatecapacity if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1606
                                                Largest FMI       =     0.4101
DF adjustment:   Large sample                   DF:     min       =     356.53
                                                        avg       =   6,175.19
                                                        max       =  31,749.17
Model F test:       Equal FMI                   F(   9,23992.0)   =       5.33
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |   .8953716   .0298361    -3.32   0.001     .8387514     .955814
    newpolitymin1sq |   .9839979   .0068432    -2.32   0.020     .9706706    .9975082
  newincumbpowerdur |   1.042535   .0186393     2.33   0.020     1.006631    1.079718
        newgdppcthl |   .9585485   .0904094    -0.45   0.654     .7966624    1.153331
          newlnoill |    .892441   .0333161    -3.05   0.002      .829472    .9601902
newmilexpsold10tile |   1.603566    .165206     4.58   0.000     1.310189    1.962636
           civilwar |   .7440531   .5432336    -0.40   0.686     .1777224    3.115055
      newcivxmilexp |   .8026153   .0995544    -1.77   0.076     .6293162    1.023637
   newstatecapacity |   .3964932   .1533808    -2.39   0.017     .1852817    .8484749
              _cons |   .0811142   .0658385    -3.09   0.002     .0164797    .3992496
-------------------------------------------------------------------------------------

. *       RESULT:  Significant at the .05 level.  Turns GDP per capita statistically insignificant
. 
. * Multiple imputation, impact on combined reduced model
. clear

. use revolutionaryeps.dta

. drop if colony==1
(57 observations deleted)

. mi set wide

. mi xtset, clear

. mi stset, clear

. mi register imputed newpolitymin1 newpolitymin1sq newgdppcthl newlnoill newincumbpowerdur newmilexpsold10tile ne
> wcivxmilexp lnparticnum deathtile10 urbxdeathtile10 newstatecapacity

. tab _mi_miss

   _mi_miss |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |        135       46.88       46.88
          1 |        153       53.13      100.00
------------+-----------------------------------
      Total |        288      100.00

. * Impute
. mi impute chained (pmm, knn(3)) newpolitymin1 newpolitymin1sq (truncreg, ll(0)) newgdppcthl (pmm, knn(3)) newlno
> ill (pmm, knn(3)) newincumbpowerdur (truncreg, ll(0) ul(10)) newmilexpsold10tile (pmm, knn(3)) newcivxmilexp (pm
> m, knn(3)) lnparticnum (truncreg, ll(0) ul(10)) deathtile10 (pmm, knn(3)) urbxdeathtile10 (pmm, knn(3)) newstate
> capacity = success civilwar urbandum democrat antimonarch, add(60) rseed(1234) force dots

Conditional models:
    newincumbpow~r: pmm newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , knn(3)
         newlnoill: pmm newlnoill newincumbpowerdur newgdppcthl newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , knn(3)
       newgdppcthl: truncreg newgdppcthl newincumbpowerdur newlnoill newpolitymin1 newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , ll(0)
     newpolitymin1: pmm newpolitymin1 newincumbpowerdur newlnoill newgdppcthl newpolitymin1sq deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , knn(3)
    newpolitymin~q: pmm newpolitymin1sq newincumbpowerdur newlnoill newgdppcthl newpolitymin1 deathtile10
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , knn(3)
       deathtile10: truncreg deathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , ll(0) ul(10)
    urbxdeathti~10: pmm urbxdeathtile10 newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 lnparticnum newmilexpsold10tile newcivxmilexp newstatecapacity success civilwar
                     urbandum democrat antimonarch , knn(3)
       lnparticnum: pmm lnparticnum newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 newmilexpsold10tile newcivxmilexp newstatecapacity success
                     civilwar urbandum democrat antimonarch , knn(3)
    newmilexpsol~e: truncreg newmilexpsold10tile newincumbpowerdur newlnoill newgdppcthl newpolitymin1
                     newpolitymin1sq deathtile10 urbxdeathtile10 lnparticnum newcivxmilexp newstatecapacity
                     success civilwar urbandum democrat antimonarch , ll(0) ul(10)
     newcivxmilexp: pmm newcivxmilexp newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newstatecapacity success
                     civilwar urbandum democrat antimonarch , knn(3)
    newstatecapa~y: pmm newstatecapacity newincumbpowerdur newlnoill newgdppcthl newpolitymin1 newpolitymin1sq
                     deathtile10 urbxdeathtile10 lnparticnum newmilexpsold10tile newcivxmilexp success civilwar
                     urbandum democrat antimonarch , knn(3)

Performing chained iterations:
  imputing m=1 through m=60 .........10.........20.........30.........40.........50.........60 done

Multivariate imputation                     Imputations =       60
Chained equations                                 added =       60
Imputed: m=1 through m=60                       updated =        0

Initialization: monotone                     Iterations =      600
                                                burn-in =       10

     newpolitymin1: predictive mean matching
    newpolitymin~q: predictive mean matching
       newgdppcthl: truncated regression
         newlnoill: predictive mean matching
    newincumbpow~r: predictive mean matching
    newmilexpsol~e: truncated regression
     newcivxmilexp: predictive mean matching
       lnparticnum: predictive mean matching
       deathtile10: truncated regression
    urbxdeathti~10: predictive mean matching
    newstatecapa~y: predictive mean matching

------------------------------------------------------------------
                   |               Observations per m             
                   |----------------------------------------------
          Variable |   Complete   Incomplete   Imputed |     Total
-------------------+-----------------------------------+----------
     newpolitymin1 |        276           12        12 |       288
    newpolitymin~q |        276           12        12 |       288
       newgdppcthl |        283            5         5 |       288
         newlnoill |        285            3         3 |       288
    newincumbpow~r |        286            2         2 |       288
    newmilexpsol~e |        240           48        48 |       288
     newcivxmilexp |        240           48        48 |       288
       lnparticnum |        267           21        21 |       288
       deathtile10 |        275           13        13 |       288
    urbxdeathti~10 |        275           13        13 |       288
    newstatecapa~y |        155          133       133 |       288
------------------------------------------------------------------
(complete + incomplete = total; imputed is the minimum across m
 of the number of filled-in observations.)

. mi estimate, post dots eform saving(miest, replace): logit success lnparticnum urbandum deathtile10 urbxdeathtil
> e10  newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile newstatecapacity 
> if startyear>1899

Imputations (60):
  .........10.........20.........30.........40.........50.........60 done

Multiple-imputation estimates                   Imputations       =         60
Logistic regression                             Number of obs     =        288
                                                Average RVI       =     0.1459
                                                Largest FMI       =     0.4263
DF adjustment:   Large sample                   DF:     min       =     330.02
                                                        avg       =   8,917.29
                                                        max       =  28,514.11
Model F test:       Equal FMI                   F(  11,37040.3)   =       5.33
Within VCE type:          OIM                   Prob > F          =     0.0000

-------------------------------------------------------------------------------------
            success |     exp(b)   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
        lnparticnum |   1.852497    .253481     4.51   0.000     1.416614    2.422498
           urbandum |   30.13947   40.61418     2.53   0.011     2.148044    422.8907
        deathtile10 |   1.214546   .1867612     1.26   0.206     .8985018    1.641758
    urbxdeathtile10 |   .5904243   .1044647    -2.98   0.003     .4173948    .8351827
      newpolitymin1 |   .8859748   .0326411    -3.29   0.001     .8242446    .9523283
    newpolitymin1sq |   .9843443   .0072545    -2.14   0.032     .9702242      .99867
  newincumbpowerdur |   1.034218   .0210437     1.65   0.098     .9937802    1.076301
        newgdppcthl |   .8740999   .0896264    -1.31   0.190     .7148857    1.068773
          newlnoill |   .9048909   .0390633    -2.32   0.021      .831469    .9847962
newmilexpsold10tile |   1.403619   .1210344     3.93   0.000     1.185196    1.662295
   newstatecapacity |   .3481012   .1517574    -2.42   0.016     .1476563    .8206524
              _cons |   .0000201   .0000387    -5.61   0.000     4.57e-07     .000881
-------------------------------------------------------------------------------------

. *       RESULT:  Statistically significant at the .05 level; turns GDP per capita insignificant
. 
. * Complete-case sample
. clear

. use revolutionaryeps.dta

. * Sigman/Hanson state capacity index (sample significantly reduced)
. * Bivariate
. logit success newstatecapacity if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        155
                                                LR chi2(1)        =       5.01
                                                Prob > chi2       =     0.0252
Log likelihood = -100.47342                     Pseudo R2         =     0.0243

----------------------------------------------------------------------------------
         success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------+----------------------------------------------------------------
newstatecapacity |   .6275847    .134479    -2.17   0.030     .4123602    .9551421
           _cons |   .4906282   .0990979    -3.53   0.000     .3302368    .7289195
----------------------------------------------------------------------------------

. *       RESULT: Not significant 
. * State capacity, by urban/rural
. logit success i.urbandum##c.newstatecapacity if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        155
                                                LR chi2(3)        =      18.98
                                                Prob > chi2       =     0.0003
Log likelihood = -93.489027                     Pseudo R2         =     0.0922

---------------------------------------------------------------------------------------------
                    success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------------------+----------------------------------------------------------------
                   urbandum |
                       yes  |   5.833031    3.40794     3.02   0.003     1.856013     18.3319
           newstatecapacity |   .3048269   .1428693    -2.53   0.011     .1216477    .7638407
                            |
urbandum#c.newstatecapacity |
                       yes  |   1.756248   .9694325     1.02   0.308      .595293    5.181328
                            |
                      _cons |   .1353953   .0724685    -3.74   0.000     .0474252    .3865431
---------------------------------------------------------------------------------------------

. *       RESULT:  Significant at .05 level for rural revolutions, not significant for urban revolutions
. * Impact on regime-specific model
. logit success newpolitymin1 newpolitymin1sq newincumbpowerdur newgdppcthl newlnoill newmilexpsold10tile civilwar
>  newcivxmilexp newstatecapacity if startyear>1899 & colony==0, or nolog

Logistic regression                             Number of obs     =        142
                                                LR chi2(9)        =      41.93
                                                Prob > chi2       =     0.0000
Log likelihood =  -73.82499                     Pseudo R2         =     0.2212

-------------------------------------------------------------------------------------
            success | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------------+----------------------------------------------------------------
      newpolitymin1 |    .904698   .0379507    -2.39   0.017     .8332915    .9822234
    newpolitymin1sq |   .9872103   .0092595    -1.37   0.170     .9692278    1.005526
  newincumbpowerdur |   1.039528   .0247152     1.63   0.103      .992198    1.089115
        newgdppcthl |    .965519   .1070193    -0.32   0.752     .7769849      1.1998
          newlnoill |   .8765008    .042284    -2.73   0.006     .7974232    .9634204
newmilexpsold10tile |   1.384479   .2057303     2.19   0.029     1.034666    1.852563
           civilwar |   .4336975   .5202703    -0.70   0.486     .0413116    4.553041
      newcivxmilexp |   .8839755   .1630523    -0.67   0.504     .6157908    1.268958
   newstatecapacity |   .4282677   .1875348    -1.94   0.053     .1815438    1.010297
              _cons |   .2035782   .2381402    -1.36   0.174     .0205598    2.015782
-------------------------------------------------------------------------------------

. *       RESULT: Not significant 
. 
. 
. log close
      name:  <unnamed>
       log:  C:\Users\mbeissin\Desktop\Stata files for book\Robustnesstestfiles\Logfiles\robustnesstestschapter4.l
> og
  log type:  text
 closed on:  26 Jan 2022, 13:02:04
------------------------------------------------------------------------------------------------------------------
